keratin mrna structure and gene organization · keratin fnrna structure and gene organization ......
TRANSCRIPT
KERATIN fnRNA STRUCTURE AND GENE ORGANIZATION
A thesis submitted to the University of Adelaide,
south Australia, for the degree of Doctor of philosophy, by
TREVOR JOHN LOCKETT B . Sc. (IIons . )
Department of Biochemistry,
University of Adelaide,
South Australia.
February , L979.
ûr,r, ^ , /" ^/ í''\"(',¡ I ?)f
To Angela
For her distractions during our courtship and her
support during our marriage.
TABLE OF CONTENTS
SUMMARY
STATEMENT
ACKNOWLEDGMENTS
ABBREVIATIONS
CHAPTER I I NTRODUCTION
A. GENERAL INTRODUCTION
B. FEATHER DEVELOPMENT
1. Keratins
2. Ernbryonic Feather Differentiation(a) Developnent and keratinization(b) Factors affecting keratinocyte develop-
ment
(i) The role of the derrnis
(ii) Vitanin A
(iii) Hormones
(iv) Epidermal growth factor(c) DNA synthes i s and rnitos is
SOME CHARACTERISTICS OF FEATHER KERATIN nRNA
STRUCTURE OF EUKARYOTIC mRNAs AND THEIR
1. nRNA Structure
2. RNA Processing
MULTIGENE FAMILIES
Faithfully Repeated Multigene Fanilies
(a) Ribosomal genes
(b) 5S genes
(c) Histone genes
Pqg"
r_v
V
vi
t_
1
1
1
J
3
4
4
6
7
7
8
9C
D
B IOGENES I S L2
TZ
15
2l
2L
22
24
25
E
1
Page
F
(d) General conclusions from these gene
arrangement studies
2. Heterogeneous Mutligene Families
(a) Imrnunoglobulin genes
(b) Silk rnoth chorion proteins
(c) Keratin genes
AIMS OF THE PROJECT
CHAPTER II - GENERAL MATERIALS AND METHODS
A. MATERIALS
1. Proteins and Enzynes
2. Radiochemicals
3. Chenicals for Specific Procedures
(a) Electrophoresis
(b) Density gradient centrifugation(c) Complementary DNA synthesis
(d) Radioactive counting
4. Miscellaneous Materials
5. Preparation of Solutions
GENERAL METHODS
Ethanol Precipitation
Preparation of Keratin nRNA
Preparation of rRNA
Preparation of Chick DNA
Preparation of Hydroxylapatite (HAP)
Preparation of cDNA
(a) cDNA to keratin nRNA
(b) cDNA to rRNA
Radioiodination of RNA
27
28
29
30
3L
32
33
33
33
34
34
34
34
55
55
55
36
36
36
3/
37
39
39
59
40
4L
B
1
2
3
4
5
6
7
Determination of Radioactivity
Digestion of À Phage DNA with EcoR,
Restriction Endonuclease
CHAPTER III - STRUCTURE OF KERATIN nRNA
INTRODUCTION
METHODS
1. Preparation of nRNA, rRNA, cDNAs and DNA
2. RNA-cDNA Hybridizations(a) Hybridization kinetics(b) Estimation of ribosomal contarninants
in cDNA preparations
(c) Conpetition hybridizations
3. cDNA-DNA Reassociations
4. Preparation of Keratin AMV- and PolI-cDNA
Duplexes for Density Gradient Centrifu-
gation
RESULTS
1. nRNA Purity2. The Nature and Specificity of AMV- and
PolI-cDNAs
3. Molecular Weights of Keratin AMV-cDNA and
PolI -cDNA
4 . Cornpetition Hybridi zation Between Pol I -
and AMV-cDNA
5. Reassociation of Keratin AMV-cDNA and PolI-
cDNA to Chick Erythrocyte DNA
6. Reassociation of Elongated PolI-cDNA to
Chick Erythrocyte DNA
8
9
Page
42
43
A
B
44
46
46
47
47
48
48
49
50
51
51
C
53
54
55
57
58
7. Thernal Stabilities of Reassociated
Duplexes of Keratin AII{V- and PolI-cDNA
8. Buoyant Densities of Reassociated Duplexes
of Keratin AMV- and PolI-cDNAs in the
Presence of Actinomycin D
DI SCUSSION
Analysis of nRNA Preparations and cDNAs
Reassociation and Thernal Properties ofAMV-cDNA and PolI-cDNA Duplexes
Model for the Sequence Organizatíon of
Keratin nRNA
CHAPTER IV - SEPARATION OF RIBOSOMAL AND KERATIN
SEQUENCES BY PHYSICO-CHEMICAL METHODS
INTRODUCTION
METHODS
1. Thermal Elution frorn HAP
(a) In aqueous buffer(b) In buffer containing 50% fornanide
2. CsCl Equilibriurn Density Gradient Fraction-
ation
3. Detection of Ribosomal and Keratin
Sequences
RESULTS
1. Thermal Elution from HAP
2. CsCl Gradient Centrifugation
DI SCUSS I ON
Thernal Elution
CsCl Gradient Fractionation
Page
60
D
63
67
60
6Z
62
72
75
75
75
76
1
)
3
A
B
C
D
/o
77
78
78
81
85
84
89
1
2
Page
90E. APPENDIX
CHAPTER V - THE HYBRID-DENSITY GRADIENT PROCEDURE FOR
A
B
THE ANALYSIS OF GENE ARRANGEMENT
INTRODUCTION
METHODS
1. Isolation of Specific Size Classes ofSingle-stranded Chick DNA
(a) Alkaline sucrose gradient centrifugation(b) Alkaline agarose ge1 electrophoresis(c) Extraction of single-stranded DNA
from preparative alkaline agarose
ge 1s
2. Pre-hybridizatíon Conditions
3. Generation of L00% Hybrid
4. CsrS0O Centrifugation
5. Detection of Hybrids
(a) RNA'ase resistance of iodinated hybrid(b) Binding of iodinated hybrid to nitro-
ce1 1ulos e
(c) Irunobilizatíon of DNA on nitro-ce1lu1ose followed by filterhybridi zation
6. Deternination of Single-stranded Sequence
Dens ity
RESULTS
1. Isolation of Specific Size Classes of
Single-stranded Chick DNA
(a) Alkaline sucrose gradient centrifu-gation
91
94
94
94
94
96
98
99
100
100
101
101
L02
L02
103
103
c
105
z
(b) Preparative alkaline agarose ge1
electrophore s i s
Studies on the Hybrid-density Gradient
Procedure for the Analysis of Gene
Arrangement
(a) Detection of 100% hybricl and single-stranded ribosomal sequence densities
(b) Pre-hybridization and detection of
hybrid in CsrS0O gradients
(c) Comments on pre-hybridization condi-
tions(d) The relationship between density and
the proportion of the DNA rnolecule
in hybrid form
Page
105
110
110
111
LT4
D. DISCUSSION
118
TzL
1
2
Isolation of Single-stranded DNA Size
Classes
Studies on the Hybrid-density Gradient
Procedure for the Analysis of Gene
Arrangement
(a) Requirernents for the hybrid-density
gradient procedure
(b) Hybridization conditions
(c) Hybrid-density studies using differ-
ent single-stranded DNA size
clas s es
(d) Experimental shortcomings
Concluding Remarks
I2T
L23
L24
r29
131
L33
1553
Page
CHAPTER VI - CONCLUDING DISCUSS,ION
1
2
Structure of Feather Keratin nRNA
Studies on Gene Organization
139
r4z
BIBLIOGRAPHY
APPENDIX
1
SU}4MARY
1. The work described in this thesis is divided into two
sections: -
(a) A study of the organization of sequences withinpurified feather keratin nRNA.
(b) The investigation of a density
ernploying RNA: DNA hybrids , f or
arrangement of tandernly linked
gradient system,
analysing the
fanilies of genes.
2. Two keratin complementary DNAs (cDNAs) of differentlengths r^rere prepared, characterized and used to study the
organization of sequences within polysomal chick feather
keratin nRNA by the analysis of their reassociation kinetics,when annealed to a vast excess of chick erytirrocyte DNA, and
by the thernal stabiLity of the duplexes formed at different
stages during the reassociation reaction. This work has
produced the following conclusions.
(a)
(b)
Each keratin nRNA molecule carries a 150 nucleo-
tide sequence adjacent to the 3t poly(A) tract,
which is represented only once in the chick
genome.
A reiterated sequence, which is either short
(about 50 bases) and faithfully conserved or
longer and nismatching, is covalently attached
to the 3t unique sequence and located further
toward the 5r end of the nRNA.
3 The possibility of enriching high molecular Ì\Ieight
11.
chick erythrocyte DNA samples for keratin sequences by
physico-chemical rnethods for future use in the analysis of
keratin gene organization was investigated. It was found
that thernal elution from hydroxylapatite (HAP) in aqueous
buffers could produce a 2.8- fold enrichment of keratin
sequences and a very large enrichment for ribosomal sequences
at the expense of DNA integrity. Sinilar elutions in the
presence of fornarnide 1ed to decreased DNA degradation and
decreased enrichment for keratin (about 1.8-fold) and ribo-
somal (5-fo1d) sequences. Caesiun chloride density grad-
ient fractionation permitted a quantitative separation of
ribosomal and keratin sequences with the possibility of
further enrichment of keratin sequences from total chick DNA
by antibiotic (actinornycin D and netropsin sulphate) CsCl
gradient s .
4. A rnethod was developed for the preparative isolation
of relatively homogeneous single-stranded DNA size classes
from randomly sheared chick erythrocyte DNA using alkaline
agarose gels for the size fractionation. A number of dif-
ferent procedures for extracting DNA frorn agarose gels I^/ere
investigated with sirnple centrifugation of the agarose slice
and collection of the supernatant proving to be the rnost
effective.
5. A caesiurn sulphate density gradient system, €flploying
pre-hybridization of RNA to single-stranded DNA, for the
analysis of tandemly repeated gene farnilies, was evaluated
using the chick ribosomal cistrons as a model system. By
using single-stranded DNA size classes in the pre-hybridiza-
tion reaction and examining the relationship between the
111 .
density of the generated hybrid (relative to the densities
of single-stranded ribosonal sequences and 100? hybrid) and
the proportion of the ribosomal sequences expected to be inhybrid form, it was concluded that the method was insuffi-ciently accurate to be generally applicable to the analysis
of the arrangement of nulti-.gene farnilies.
r_v.
STATEMENT
This thesis contains no rnaterial which has been
accepted for the award of any other degree or diplona inany university. To the best of my knowledge, it contains
no material that has been previously published by any other
person, except where due reference is made in the text.
Signed
Trevor J. Lockett
I would likenission to work inof Adelaide.
V
ACKNOWLEDGMENTS
to thank Professor W.H. Elliott for per-
th-e Department of Biochemistry, University
I arn nuch indebted to Professor G.E. Rogers for hisenthusiastic supervision, advice and encouragement through-
out the course of this work.
Thanks are also due to ny colleagues in the featherproject for their friendship and stirnulating discussions
Messrs. R.B. Saint, B.C. Powe11 and C.P. Morris, who were
present throughout the course of this project and Drs. D.J. Kenp
and P.E.M. Gibbs who were present during the fornative years.
rn particular, my thanks go to Dr. Kernp for his collaborationwith some experiments described in Chapter III, and for per-
mission to quote some of his results.
I am grateful to Mrs. Lesley Crocker for her excellenttechnical assistance and invaluable help in preparing and
photographing diagrams. Thanks are also due to l,{iss P.Y. Dyer
for assistance with electron microscopy and useful discussions.
I also thank Mrs. J. Somerville fol typing the thesis.
f was supported throughout this work by a Commonwealth
Postgraduate Research Award. The work was supported by
grants from the Australian Research Grants Connittee and
the Australian Wool Corporation (to G.E. Rogers).
ABBREVIATIONS
The following abhreviations have been used in thisthesis.
v]- .
bovine serum albumin.
curle.
synthetic DNA complementary to messenger RNA.
caesiun chloride.
caesiun sulphate.
deoxyribonucleic acid.
ethylenedianinetetracetic acid.
hydroxylapatite.
heterogeneous nuclear RNA.
kilobase (RNA or single-stranded DNA)
kilobase pairs (Native DNA).
messenger ribonucleic acid.
oligodeoxythymidylic acid.
E. coli DNA polymerase I.polyadenylic acid.
ribonucleic acid.
ribonuclease.
messenger ribonucleoprotein complex.
ribosomal RNA.
sodiun dodecyl sulphate.
standard saline citrate (0.15 lr{ NaCl, 0.015
N{ Na citrate).
saturated solution of potassium iodide.
trichloroacetic acid.
10 nM Tris-HCl pH 7.5, 1 mM EDTA.
temperature at which half the nolecules in
B. S .A.
cDNA:
CsCl :
CS TSOO
DNA:
EDTA:
HAP:
hnRNA:
kb:
nRNA:
oligo dT:
PolI:
Poly (A) :
RNA:
RNA I ase:
nRNP:
rRNA:
SDS:
SSC:
SSKI:
TCA:
T.E.:
T:m
vt].
IRNA:
the DNA duplex or RNA:DNA hybrid under study
are denatured.
transfer RNA.
CHAPTER I
INTRODUCTION
1
CHAPTER I
INTRODUCTTON
A GENERAL INTRODUCTION
The enbryonic chick down feather constitutes an
exanple of a terminally differentiating tissue. After an
initial period of growth, the cells become filled with the
fibrous protein, keratin, and eventually die. The general
ain of the work in this laboratory is to gain an understand-
ing of the molecular events leading to a feather ce1lfs
commitment to keratin production. The work described in
this thesis falls into two parts. The first has been to
study the nature of different regions of the keratin nRNA
mo1ecule. The second has been directed towards establish-
ing a density gradient rnethod for analysing gene organizati-on
with the intention of applying it to a study of the arrange-
ment of keratin genes in the chick genome.
The ensuing literature survey is intended to provide
relevant background information on feather development in
addition to information on current views concerning nRNA
structure and eukaryote gene organizatior..
B FEATHER DEVELOPMENT
1 Keratins
Keratins are fibrous, insoluble, intracellular
proteins which have a high cystine content. They are found
in the epidernis and tissues derived from it, ê.9., the
feathers, claws, beaks and scales in birds and the claws and
hair in animals. Within the cells, the proteins are
2
aggregated into fibrillar masses, the structure of which has
been extensively studied (see Fraser et al., 1972, for re-
view). Owing to the stabiLization of these protein aggre-
gates by disulphide bonding (Goddard and Michaelis, 1934) ,
the study of the proteins requires disulphide bond breakage
by reducing agents, followed by stabilization by alkylation,
e.g., carboxymethylation (Harrap and Woods, I964a; Kemp and
Rogers , I972).
S-carboxynethylated chick feather keratins appear to
be homogeneous in nolecular weight, the estimates varying
between 10,500 and 11r500 daltons (Harrap and Woods, 1964b;
Walker and Rogers, I976a), while reduced, non-alkylated
keratin had a molecular weight of 10,000 daltons (Woodin,
1954) . Despite this molecular weight homogeneity, the
proteins appear to have different charge characteristics,
as determined by ion exchange chromatography and electro-
phoresis, even when the material is obtained from one parti-
cular keratinized tissue (Harrap and Woods, I964a; Kernp and
Rogers, 1972; Walker and Rogers, L976a). Some ten electro-
phoretically distinguishable protein species from the rachis
component of adult feathers were resolved by Woods (1971),
while Walker and Rogers (I976a) have found a ninimum of
twenty-two protein chains in the embryonic feather. Sequence
analysis of the proteins has suggested that each chain is
the product of a separate gene, but that all the proteins
are closely related in prirnary structure (Walker and Rogers,
1e76b).
Although embryonic feathers, and the specific parts
of the adult feather (Rawles, 1965) , contain an assernbly of
3
closely related protein chains, each tissue has its own
distinct population of protein species (Harrap and Woods,
I964a; Kemp and Rogers, L972). Furthermore, the proteins
of feather exhibit fundanental differences from those of
other keratinized tissues, such as scale (Kenp and Rogers,
L972; Walker and Bridg€n, L976).
Despite these differences, avian keratins appear to
have evolved from common ancestraL proteins. The spectrum
of proteins observed in different species of birds reveal
similar degrees of heterogeneity and tissue specificity,
although the patterns for a given tissue show some inter-
species variation (Harrap and Woods, 1967; Woods, I97L;
OfDonnell, L973a). Prirnary structure analysis of chains
fron the adult feather calarnus of emu (0'Donnell, I973b) and
gul1 (0'Donnell and Inglis, L974), from chick embryonic
feather (Walker and Rogers, 1976b) and fron chick scale
(Walker and Bridgêr, I976) illustrates the presence of major
regions of sequence homology between the proteins from dif-
ferent tissues.
2. Enbryonic Feather Differentiation
(a) Developrnent and keratinizat:.on
The cellular events involved in ernbryonic
feather morphogenesis have been extensively studied
and detailed reviews can be found elsewhere (Watterson,
L942; Romanoff, 1960; Li11ie, 1965; Voitkevich, 1966;
Matulionis , 1970). Since keratinization is the nani-
festation of terminal differentíation, this survey
will only deal with events occurring during and sub-
4
sequent to the onset of keratin synthesis.
Keratin synthesis begins at about the
twelfth day of embryonic development, as judged by
the appearance of keratin fibrils in the cells (Matu-
lionis, 1970) and, âs shown in Figure 1.1, by the
appearance of keratin proteins on acrylanide gels of
S-carboxynethylated proteins extracted from feathers
of different ages. Figure 1.1 shows that bands
corresponding to the rnajor feather keratin proteins
rtrere only present in trace amounts in 11- and lZ-day
feather extracts. These bands rapidly increased in
quantity after day LZ and had become the most abundant
protein species in the feathers by day 15 (Kernp et al.,
L974a). It hras also observed, from labe11ing studies
and ge1 electrophoresis of proteins, that the keratin
proteins are co-ordinately synthesized in the tissue
(Kernp et al. , I974a). Keratinization of the feather
is completed by about the eighteenth day of embryonic
development and the chick hatches at 2l days.
(b) Factors affecting keratinocyte development
A number of factors have been shown to
affect the growth and development of keratinizing
tissues and these are briefly discussed below. A
more extensive review may be found in Fraser et aL.,
(Le72).
(i) The role of the dernis
When the dernis
taken from different regions
and epidermis are
of an embryo,
FIGURE 1.1.
TIME COURSE OF KERATIN SYNTHESIS
Sanples (100 uglge1) of the reduced and carboxymethyl-
ated protein preparations from feathers at 11-19 days of
embryonic feather develop¡nent were subjected to pol"yacrylarnide ge1
electrophoresis at pH 9.5 and stained with Coomassie blue.
O, origin; + anode; o,, ß and Y bands, keratin.
Reproduced fron Kernp et al. , L974a.
ÇI
o
lF
]"
I4,32I
p
5
4
tat
+11 12 13 14 15 19
DAYS
5
recombined and allowed to develop in tissue cu1-
ture, the epidernis develops structures character-
istic of those nornally associated with the
dermis used (Senge1, L957, 1-971-; Rawles, 1963,
1965; Wesse11s, 1962, 19ó5). Furtherrnore, the
keratins produced in these recombinants are
identical to those found in the same epiderrnal
structures where no recombination has been per-
forned (Dhouailly et dL.,1978). Thus the re-
combination of presumptive scale dernis with
presumptive feather epidernis causes the epider-
mis to form scales, and the keratins present in
these scales are the same as those present in
the nornal tissue. These effects, exerted by
the dermis, only occur at specific stages of
development (Rawles, 1965, 1965) which are much
earlier than the onset of keratínization in these
tissues (Matulionis , L97 0; Kernp et àI. , L97 4a;
Beckinghan Snith, 197 3a) .
While the rnajor role of the dernis
appears to be to control the nitosis of epider-
mal ce11s, thereby controlling epidermal mor-
phology (Wesse1ls,1962; Dodson, 1965), it is
clear that it also specifies the genes which
are being expressed in the differentiating
tissue. It would appear that the control by
the derrnis over the epiderrnis is most likely
mediated by some extracellular substance from
the mesoderm influencing the environment of
epidermal ce1ls, since Wessells (1962) has shown
6.
that isolated epiderrnis would gror^r even if a
nillipore filter was interposed between the
dermis and epidermis prior to culturing the
whole assembly in vitro. In studies using het-
erospecific dermal-edpiderrnal recombinants,
Dhouailly (1967) has shown that duck derrnis can
direct chicken epiderrnis to produce feathers
resembling those of the duck. In similarexperiments (Dhouailly et â1., 1978), mouse foot-pad dermis was shown to direct the production ofa footpad structure in presumptive feather pro-
ducing chick epiderrnis. The keratins of these
footpad structures were shown to be of the scale
type. These results suggest that the effectorof dermal control over the epiderrnis is not
species specific, and even in heterospecific
recombinants, it can dictate the type of gene
to be expressed.
(ii) Vitanin A
It has been shown that if immature
(Fe11 and Me11anby, 1953) or highly differen-
tiated (Fe11, 1957) ernbryonic chick skin is
cultured in the presence of excess Vitamin A,
keratinization is inhibited and the cel1s under-
go a mucous metaplasia. The synthesis of kera-
tin proteins is completely repressed under these
conditions (Beckinghan Smith, 1973b). For the
older tissue at least, the effects are revers-
ible on the removal of vitamin A (Fitton-
7
Jackson and Fe11, 1965) .
(iii) Horrnones
Several hormones have been impli-
cated in the developnent of avian keratinocytes.
The role of pituitary and thyroid glands has
been discussed at some length by Voitkevich
(1966), but few studies of these phenomena have
been made at the molecular level.
In vivo (Bartels , 1943) and in
vitro experiments (Wessel1s, 1961; Kitano and
Kuroda, 1967) have indicated that thyroxine
accelerates epidermal keratinization. In the
light of this evidence, it is interesting to
note that the thyroid gland reaches maxirnum
thyroxine secretion (Shain et al. , 1972) at
about the same tine as the onset of keratin
synthes is .
Hydrocortisone also hastens kera-
tin synthesis in skin cultures (Fe11, L962;
Suginoto and Endo, 1969) and causes feather
germs to abort. The presence of hydrocortisone
in the medium was found to stinulate the syn-
thesis of one class of scale epidernal proteins,
although another class is not synthesised
(Sugimoto et aI., L974).
Epidermal growth factor(iv)
This polypeptide growth factor was
8.
first isolated fron the submaxilliary gland ofthe nale mouse by Cohen (L962) and produced pre-
cocious epiderrnal development when adninistered
to in vivo or in vitro organ culture (Cohen,
1965). The growth factor stimulated RNA and
protein synthesis (Hoober and Cohen, 1967) and
an accompanying increase in the proportion of
the ribosomes associated with polysones (Cohen
and Stastny, 1968) . The material isolated fron
mouse is effective against skin tissue fron a
variety of animal sources, including the chicken,
and sinilar material has recently been isolated
from man (Cohen and Carpenter, 1975), suggesting
that it might be widely distributed arnongst
animals. An in vivo requirement for epidernal
growth factor has not, however, been denonstrated.
(c) DNA synthesis and rnitosis
From the above discussion it is apparent
that many factors affect keratini zation and that there
can be no simple mechanism whereby keratin synthesis
is initiated in feather or skin cells. It is note-
worthy that many of the factors discussed appear to
affect mitosis, and that some effects on differen-
tiation (or pseudo-differentiation in the case of
vitarnin A) are observed in post-nitotic cel1 popula-
tions. It has been shown that DNA synthesis in skin
(Rothberg and Ekel , L97T) and DNA polyrnerase activityin feather (Kischer and Furlong, 1967) are at their
maximum levels inmediately prior to the onset of
9
C
keratinization in the tissues, and decrease signifi-
cantly by the tine keratin synthesis is established.
It is possible that the cells are commit-
ted to keratin synthesis earLy in development, as has
been observed for rnyogenic cel1s (e. g. , Holtzer et ãL. ,
1972, 1973). These cells, although not actively syn-
thes izing muscle proteins, undergo several rounds of
DNA synthesis and mitosis, until after a "quantal"
division, the cells become ful1y differentiated.
Depending on the environment in which they are located,
epidermal cells show a certain degree of developnental
variability and perturbations of that environment can
cause the cells to follow a different developmental
pathway. Nevertheless, the cel1s could be initially
constrained to follow one of a limited number of fates,
and the association with a particular mesenchyme could
lead to the ce11s being finally committed to one path.
Thus, with the appropriate dermal stimulus, the cells
could undergo a programmed pattern of cel1 division,
culminating in a "quantal" mitosis and keratinization.
After the quantal division, the cells would be in-
capable of responding to external stimuli; however
prior to ít, najor changes in environment could lead
to a different set of genes being expressed, âS, for
example, in the case of the mucous netaplasia induced
by vitanin A (Fell, 1957).
SOME CHARACTERISTICS OF FEATHER KERATIN NRNA
The
aI. , L973;
iso 1at ion
Kemp et
of feather
ãL., f974b)
keratin nRNA (Partington et
has greatly aided studies on
10.
feather keratin synthesis. The nRNA is isolated from L4-
day embryonic feathers and, when purified, sediments at
about 12S in sucrose gradients (Partington et a1., L973;
Kenp et al. , L974b). By the comparison of the nobilitiesof feather keratin nRNA to those of 28S and 18S rRNA, 55
RNA and tRNA on acrylamide gels in the presence of 98% for-mamide, the nolecular weight of the nRNA was calculated tobe 2.5 x 105 daltons or 760 nucleotides (Kernp et ãL. , Lg74b) .
The nRNA stimulates keratin synthesis in both the rabbitreticulocyte (Partington et ãT., L973) and wheat embryo cel1
free translation systens (Kernp et ãI., L974c). In the
wheat enbryo system, the only proteins synthesized r4rere ker-
atins. From these translation studies and the observation
that purified keratin nRNA electrophoresed as a single band
on formanide gel electrophoresis (Kemp et ã1., 1974b), the
nRNA preparations r^rere judged to be about 95% pure.
Keratin nRNA was found to bind to cellulose (Parting-
ton et aI., 1973; Kemp et ãL., I974b) suggesting that it
contains a 3t poly(A) tract (Kitos et aT., L972; Schultz
et ãL., 1972; Delarco and Guroff, 1973). This is a feature
common to most other messenger RNAs studied (see Brawermann
(L974) for review) with the notable exception of histone
nRNA (Adesnik and Darnell, 1972) . The average length of
the poly (A) tract at the 3t end of feather keratin nRNA is
60 nucleotides (Morris and Rogers, in press) .
Many animal (Ad
L975a; Perry et al.,
1975b; Wei and lt4oss,
ams and Cory , L975; Furuichi et a1. ,
1975) and viral mRNAs (Furuichi et ãI.,
1975) have been shown to possess a
structure having the sequence rTG5'nnn5'xtpY(m)pZp---,
11.
rh"te r7G is 7-rnethylguanosine, which is coupled. via a tri-
phosphate linkage to the next nucleotide in the RNA chain
(X) , which bears a 2t -O-rnethyl group. The third nucleotide
(y) is also occasionally nethylated at the 2r-position.
Evidence has recently been obtained suggesting the presence
of 7-methylguanosine at the 5r terminus of keratin nRNA
coupled by a triphosphate linkage to the penultinate residue
(Morris and Rogers, in press). The sequence of the adja-
cent residues could not be determined, as it was necessary
to label the nRNA in vitro in order to detect the 5r-termin-
aL structure (c.f. Symons, 1975).
The hybridization kinetics of keratin nRNA with cDNA,
prepared using the RNA dependent DNA polymerase of avian
myeloblastosis virus, indicated that keratin nRNA is a
heterogeneous mixture of species (Kernp, 1975). From this
hybridization analysis it was estirnated that there were 25
to 35 keratin nRNA species, âr estimate consistent with the
number of protein chains in the tissue (Walker and Rogers,
L976a). Each nRNA species appears to contain a sequence
which is unique to it, and a sequence hornologous to the other
rnRNAs in the mixture. Since the rnRNA is about 760 bases
long and feather keratin chains are only 100 amino acids in
length, only 300 of the 760 bases can be attributed to the
coding sequence, the rest presumably being untranslated
sequences. Early reassociation kinetic analysis suggested
that there are between 100 and 240 genes coding for keratin
in the chick genome (Kemp, 1975).
Keratin nRNA, therefore, shows
to most eukaryotic nRNAs. However,
many features common
the sequence complexity
rz.
indicates that keratin rnRNA is a mixture of closely related
sequences, presumably exhibiting equivalent heterogeneity
to that observed in the keratin proteins.
D STRUCTURE OF EUKARYOTIC mRNAs AND THEIR BIOGENESIS
The following discussion focuses attention on the
relationship between the coding and untranslated sequences
of a number of eukaryotic nRNA species and the nechanisn by
which the final nRNA species are generated.
1 nRNA Structure
All eukaryotic nRNAs which have been character-
ized in detail are known to contain more bases than is
necessary for coding, in addition to poly(A) (Gould and
Hanlyn, 1973; Berns et a1., L972; Brownlee et â1., L973;
Milstein et al., 1974; McReynolds et al., 1978). With the
advent of rapid DNA sequencing techniques (Maxan and Gilbert,
L977; Sanger and Coulson, I975; Brownlee and Cartwright,
L977) and cDNA cloning (Maniatis et a1., 1976), rapid pro-
gress has been made in the understanding of rnRNA structures.
The most widely studied messengers have been
those coding for globins. Rabbit B globin mRNA was the
first nRNA to be completely sequenced (Baralle, 1977a;
Proudfoot, L976a, L976b; Efstratiadis et ãL., L977). Its
length, not including poly(A), is 589 bases; the 3t non-
coding sequence is 95 residues long (Proudfoot, L976a) and
the 5r non-coding sequence is 53 residues long (Baralle,
L977a). Human ß globin has also been completely sequenced
(Proudfoot, L977; Marotta et aI., L977; Baralle, L977c).
Its length,minus poly(A), is 626 bases (Baral1e, I977c), L34
L3.
of these being attributable to the 3r non-coding sequence
(Proudfoot, L977), and 50 to the 5r non-coding sequence
(Baralle, L977c) .
Comparison of the prirnary structure of rabbit
and human ß globin nRNA sequences reveals that, ignoring
the 39 base insert which is present in the 3t non-coding
region of the human, but not the rabbit, nRNA, the 3t non-
coding sequences are 77% homologous (Proudfoot, L977) and
account for about 2/S of all the non-coding sequences in
both of the nRNAs. The 5r non-coding region of rabbit ß
globin nRNA is 5 nucleotides longer than that for human but
they show 80% homology (Baral1e, L977a; Baral1e, T977c).
The homology of these non-coding sequences in these two
mamrnals suggests that these sequences are under moderate
selective pressure; less than those nucleotides specifying
amino acid sequence but more than those whose function may
be regarded as not being sequence specific, for example,
satellite DNA (Southern, 1975a).
Large portions of human (Forget et aI., L974;
Proudfoot and LongIey, L976; Proudfoot et ãI.,1977;
Baralle, L977c) and rabbit (Proudfoot, L976a; Baralle, I977b;
Proudfoot et ãI., L977) o¿ globin nRNAs have also been
sequenced. Their lengths, minus poly(A), are 575 and 551
nucleotides respectively. Human o globin nRNA has a 24
base insert in the 3t non-coding sequence which is not pre-
sent in the rabbit messenger. Ignoring this, these se-
quences ar e 80% homologous (Proudfoot et aL., L977). The
5 I non-coding sequences for human and rabbit cl globin nRNA
are 37 and 36 nucleotides respectively and exhibit 8I%
L4.
homology (Baralle,
of the nucleotides
I977c). As for the ß globins, about r,4
of the o globin nRNAs are untranslated
being at the 3t end.wirh z/s of these
Chick ovalburnin nRNA has also been cornpletely
sequenced (McReynolds et ãL.r 1978; Cheng et a1., 1977;
Brownlee and Cartwright, 1976) and shares some interestingstructural sinilarities with the globin nRNAs. The messen-
ger is 1859 residues long, not including poly(A) and 5r cap
structures. This is made up of a 673 base 3t non-coding
sequence, a 1158 nucleotide sequence coding for the ovalbu-
nin protein , and a 5 I untranslated sequence of 64 residues.
It is interesting to note that the 5 ' non-coding sequences
of human and rabbit o and ß globin nRNAs and chick ovalbu-
nin nRNA are all of similar length. There is, however, no
sequence homology between the 5 I untranslated sequences of
ovalbumin and globin mRNAs (McReynolds et al., 1978). This
lack of honology has led to the speculation that the 5 ' cap
structure and the initiating AUG codon are the only irnpor-
tant general protein synthesis initiating signals in
higher organisms (Baralle and Brownlee , 1977) . I4any authors have
also alluded to the possibility of irnportant interactions
between 1BS rRNA and sequences in this 5r region sinilar to
those proposed to occur between E. coli 165 rRNA and sequences
at the 5 I end of a number of procaryotic nRNAs (Shine and
Dalgarno, 1975) .
The consistent discovery of long untranslated 3t
terminal sequences in many different nRNAs (lvlcReynolds et
ãL., 1978; Proudfoot et â1., L977; Proudfoot, I977; Harnlyn
et ãL., T977) suggests that these sequences may have an
15.
important role in regulating nRNA function (Proudfoot et ã1, ,
L977). Baralle (I977a) has proposed that the 5r and 3t
non-coding sequences of rabbit ß globin nRNA could interactvia the 5 t U-C-C-C-C 3t sequence between nucleotides 36 and
40 and the 5r G-G-G-G-A 3t sequence between residues 525-
529. He suggests that , by such an interaction, the 3t non-
coding sequence could be involved in the initiation of pro-
tein synthesis. Alternatively the 3t non-coding sequence
night be irnportant for the regulatory functions relating to
nRNA processing frorn its presumptive precursor, transport
across the nuclear membrane or stability (Perry, 1976).
Certainly the conservation of the sequence 5 I A-A-U-A-A-A 3t
at a site about 20 nucleotides from the 3t poly(A) tract in
rabbit (Proudfoot, L976a) and human o and ß globin nRNAs
(Proudfoot et aL., L977; Proudfoot, I977), mouse immuno-
globulin nRNA (Milstein et ãI., L974) and chick ovalbumin
nRNA (Cheng et ãI. , L976) is suggestive of some important
general function for at least this portion of the sequence.
At present, however, the assignment of functions to differ-
ent portions of the untranslated sequences in the differ-
ent messengers studied can only be considered to be specula-
tive. While it is often possible to draw secondary struc-
tures for the sequenced nRNAs on the basis of base pairing,
the unknown effects of the proteins, associated with the
mRNAs in vivo, oû secondary structure, render it difficult
to make aîy firn conclusions about the existence of such
postulated structures .
RNA Processing2
Several types of structural and kinetic evidence
16.
indicate that heterogeneous nuclear RNA (hnRNA) contains
nRNA precursors in a variety of forms (see Perry et ãL. ,
1-976 for review). This section, however, will concentrate
on recent results, obtained from restriction enzyme analysis
and recombinant DNA technique, which demonstrate conclusive-
Iy the existence of nRNA precursors and the intricacies ofRNA processing.
The first strong evidence that nRNA biogenesis
need not necessarily be a sinple event, came fron the
adenovirus system, where transcripts from several parts of
the adenovirus genome rtrere found covalently linked together
in a single rnRNA molecule (Berget, êt aL., L977; Chow et aL.,
L977). These results suggested that one or more of a
number of mechanisms of rnRNA biogenesis could be acting
(KlessiB, 1-977).
(a)
(b)
(c)
RNA could be transcribed from each siteand subsequently ligated.
RNA polymerase could skip the sequence
between those regions which constitute
the nRNA (intervening sequences) forming
an intact nRNA from the one transcrip-
tional event.
RNA polynerase could copy the entire seg-
ment of DNA into an RNA product, with the
portions not required in the nRNA being
deleted, and the RNA re-1igated.
sequence ln
and F1ave11,
With the discovery of a 600 base intervening
the rabbit ß globin structural gene (Jeffereys
I977), it becorne apparent that a complex
17.
mechanisrn for the synthesis of rnature nRNA species in eukary-
otes night be a conmon phenomenon. It is interesting to
note that the rnouse ß globin nRNA precursor described by
Curtis and Weissmann Q976) and Bastos and Aviv (L977), is
only slightly longer than an hypothetical precursor nRNA
containing the rabbit ß globin coding sequences, interven-
ing sequences and 3t poly(A) tract. This observation sug-
gests that the mechanism of nRNA biogenesis most likely to
be acting in the maturation of rabbit ß globin nRNA, is that
described in (c) above.
Intervening sequences have been found in the
genes coding for mouse ß globin, mouse immunoglobulin light
chains (both rc and À), chick ovalbumin, SV40 large T anti-
gên, adenovirus nRNA and a subset of the genes coding for
the 28S rRNA of Drosophila rnelanogaster. The details of
these are outlined in Table 1.1. The arrangement of inter-
vening sequences in the chick ovalbumin gene is particularly
complex and is worthy of further discussion.
The presence of intervening sequences in the
ovalbunin gene rnras first reported by Breathnach et ãI.,
(Ig77). Using restriction enzyme analysis of total chick
DNA in conjunction with the Southern transfer procedure
(Southern, 1975b), the natural ovalburnin gene rlilas shown to
contain at least two intervening sequences of lengths 1.0
and 1.5 kb (Breathnach et 4L-, L977; Lai et al., 1978;
Doel et al. , 1977) . At least one of these intervening
Sequences was shown to interrupt the coding Sequence and
there was no difference in the sequence organization
observed in ovalbumin producing and non-producing cells
TABLE 1.1
Gene No. ofIntervenilgSequences
Size ofInterveningSequences
References
Rabbit ß globin
It4cuse ß globin
Chick ovalbr-unin
It4cuse inrnr.rroglobulin
À light chain
(lt4yelona Dl,lA)
lr4cuse inrnrnoglobulin
< light chain
(NAyelona DlttrA)
SV40 late large T
antigen
Adenovirus nRNIAS
Drosophila 28S rR},IA
1
2
600 bases
550 bases
<125 bases
200 to
1,600 bases
95 bases
L1250 bases
Jeffereys fi Flavell,
(re77) .
Tilghman et a1. (1978a)
Kinniburgh et a1. (1978).
tvfandel et a1. (1978)
Brack Ç Tonegawa Q977).
7
78(1eet algawaTone')
78)9! 4'(rgczykDugai
)
>.I
280 bases
5,000 bases
Rabbitts & Forster, (1978)
tvlatthyssens Q Tonegawa
(le78) .
Lavi Ë Groner (L977)
Bratosin et al.(1978)
Berk Ê Sharp (1978).
Chow et al. Q977) .
Glover Q Hogness, (1977)
Wtrite Q Hogness ,(L977)
Pellegrini et a1. (1977).
1
1
1
Nunber and length of intervening sequences l-n a
number of eukaryotic structural genes.
18.
(Weinstock et ã1., 1978). These and later results suggested
that intervening sequences are not used to switch trans-
cription of the ovalbunin gene on or off in the various cell
types (tr{andel et ãI., 1978). Thus, while the arrangement
of innunoglobulin variable and constant region genes have
been shown to vary for DNA derived fron germline and myeloma
DNA (Hozuni and Tonegawa, 1976; Rabbitts and Forster, 1978),
Weinstockrs results suggest that such gene re-arrangements
are not a general prerequisite for gene expression (Wein-
stock et al., 1978). When a cloned genomic ovalbumin gene
became available, it was possible to demonstrate that the
actual arrangement of intervening Sequences hIaS even more
complex (Mandel et a1., 1978; Dugaiczyk et al., 1978) with
seven such sequences all being confined to the region
corresponding to the 5r half of the ovalburnin nRNA. The
inserts varied in length from 0.2 to 1.6 kb and the eight
coding fragments rltlere shown to be arranged in tbe Same order
and relative orientation as in the nRNA (Mandel et al.,1978;
Dugaiczyk.et al. , 1978) . In addition, studies on the oval-
bunin gene intervening sequences from different chickens
indicated the existence of at least two alleles of the gene,
the variation responsible for this polyrnorphisn lying in an
intervening region (Weinstock et al., 1978; Mandel et a1.,
1e 78) .
If the entire natural ovalbumin gene is trans-
cribed in stimulated chicken oviduct ce1ls and the mature
nRNA obtained by deletion of those sequences complementary
to the intervening regions, then the pre-rnRNA would be
expected to be approximately 6 kb long (Dugaiczyk et aL.,
1978). Very recently, Roop et al., (1978) have described
19.
the extraction of nultiple species of hnRNA from stimulated
chicken oviduct ce11s, which hybridized to both structural
and intervening ovalbumin sequences. These RNA species
ranged from 1.3 to 4 tines the síze of the mature ovalbumin
nRNA. Catteral et aL. (1978) have shown the existence of
short, partiaL sequence homology at all splice points in the
chick ovalbumin gene. In the light of the existence of a
precursor for ovalbumin nRNA, it is tempting to speculate
that these homologous sequences are signals for a splicing
enzyme (Catteral et ãI., 1978). The mechanisn by which
adjacent coding sequences would be brought into close prox-
irnity for splicing was not irnnediately obvious from this
sequencing study since there was no evidence for the occur-
rence of strong Watson-Crick base pairing between adjacent
junctions.
The existence of precursor mRNAs for chick oval-
bunin ßoop et al. , 1978) , irnrnunoglobulin rc light chain
(Gilrnore-Herbert and Wal1, 1978) and mouse ß globin (Bastos
and Aviv, L977; Curtis and Weissmann , T976) , along with the
observation by Tilghrnan et a1. (1978b) that the intervening
Sequences in the natural mouse ß globin gene are trans-
cribed within the ß globin rnRNA precursor' suggest that most
genes with intervening sequences might be conpletely trans-
cribed, with deletion of intervening sequences and ligation
of coding Sequences occurring subsequently to produce the
mature mRNA. The origins and functions of intervening
sequences are unknohln. It has been postulated that these
sequences rnay be descended from some putative eukaryotic
insertion element (see Bukhari, Shapiro and Adhya, I977)
inserted at some stage during evolution to perform an
20.
unknown functi.on. Alternatively, they may represent
sequences which happened to separate regions of the genome
which nature then chose to link at the level of the messen-
ger to yield a new protein (Mandel et al., 1978). Gilbert
(I978) has suggested that intervening sequences lray a11ow
an increased rate of evolution for eukaryotic organisms.
They could also have irnportant roles in cascade type regula-
tion of gene expression at transcriptional and post-
transcriptional levels (Mandel et ãI., 1978). fn hormone-
induced systems like ovalbumin, hormone-protein complexes
may be directly involved by binding to intervening sequences
in the DNA or to their transcripts; alternatively, they may
act indirectl-y by inducing the synthesis of a specific
splicing enzyme (lvlandel et ãL. , 19 78) .
While there is good evidence that at least some
of the nRNAs derived fron genes containing intervening
sequences have high molecular weight precursors generated
by the transcription of coding plus intervening sequences,
the presence of an intervening sequence is not a prerequi-
site for a precursor:nRNA relationship. The most notable
exceptions to the above pattern of nRNA biogenesis aTe the
histone genes and nRNAs of sea urchin. The arrangernent of
these genes have been extensively characterized (Schaffner
et a1., Lg76; Gross et al., L976b; Sures et al., L976; Wu
et al., Lg76) and will be discussed in detail below. The
feature of their sequence organi zation relevant to the
present discussion, however, is that they contain no inter-
vening sequences. Nevertheless, Levy et al. (1978) have
demonstrated the existence of high molecular weight precur-
sors to each of the histone nRNAs of the sea urchin,
2I.
S. purpuratus.
E MULTIGENE FAMILIES
When evolutionarily related genes code for products
with identical, sinilar or overlapping functions and when
they are closely linked on the chromosome, they are said tobe a multigene farnily (Hood et ã1., 1975). Until the advent
of recombinant DNA technology, the study of eukaryote gene
arrangement and control was linited to those genes which
were reiterated and relatively easily purified, by physico-
chenical means, from the remaining genomic DNA. These
included the genes coding for rRNA,55 RNA (Brown et ãI.,L97L) and tRNA (Clarkson and Birnstiel, L973) in Xenopus
laevis, and those coding for histones in sea urchin (Birn-
stiel et ãL., L974). The ribosomal genes were studied in
the greatest detail (see Birnstiel, Chipchase and Spiers,
I97L, for review of this physico-chemical work). The
following survey is linited to work involving restriction
enzyme analysis and cloning, and is divided into two
sections:
Faithfully repeated multigene families.
Heterogeneous multigene fanilies.
1. Faithfully Rep eated Multigene Families
This definition refers to genes which exist in
rnultiple identical copies in the genome. The three families
that will be dealt with are the ribosomal genes of Xen opus
laevis and Drosophila melanogaster, the 55 genes of Xenopus
laevis and the histone genes of sea urchin and D
1
)
gas ter.. melano-
') ')
(a) Ribosomal genes
The ribosomal DNA of Xenopus laevis con-
sists of about 500 repeating units, each containing
a region coding for the 40S rRNA precursor and a non-
transcribed spacer region. All 500 repeating units
are tandemly arranged in the nucleolus organizer
region of one of the 18 X. laevis chromosomes.
lVhile the region transcribed into the 40S rRNA pre-
cursor is constant in all repeating units, the non-
transcribed spacer regions vary in length causing
repeat unit lengths to vary from 10.8 kb to 16.7 kb
(Wellauer et ãT., L976a). It appears that a large
fraction of the spacer region of the rDNA is composed
of short subrepeats which constitute internally repe-
titious segments. Long spacers are distinguished
from short spacers by having more copies of these
subrepeats. The observed length variation of rDNA
exists within a single nucleolus organizer and 50eo to
68% of the adjacent repeats differ in length. It
was observed, however, that the relative abundance of
the size classes contained in the chromosomal and
arnplified ribosomal cistrons from a given individual
can vary ffellbuer et al ., L976b). Furthermore, the pre-
ference for the anplification of a particular size
class is inherited and some animals selectively arnplify
repeat units which are rarely found in their chromo-
somal repertoire. Most tandem repeats in a single
anplified rDNA are of equal length (Wellauer et al.,1976b)
which supports the hypothesis that rDNA is amplified
by a rolling circle mechanism (Hourcade et ãL.,
23.
1973a, b; Rochaix g! al., 1974).
The ribosomal g enes of Drosophila me lano -
gaster also exhibit length heterogeneity. Glover
and Hogness (1977) discovered the existence of a 5 kb
insert in some structural genes coding for 28S rRNA,
and an accompanying increase in repeat unit length
from 12 kb to 17 kb. I'R-loop" mapping confirmed the
existence of the intervening sequence and "in situ"
hybridization of this sequence to D. melano aster
salivary gland chrornosomes, revealed that it was loca-
ted in the centromeric heterochrornatin and many bands
on the euchromatic arms (White and Hogness, L977) .
Subsequent studies showed that the insertions into
the gene for 28S rRNA vary from about 0.5 kb to 6.0 kb,
and occur in distinct size classes which are multiples
of 0.5 kb (Wellauer and Dawid, L977). While rDNA is
found on the X and Y chromosomes of D. melanoqaster ,
only rDNA fron the X chromosome carries 28S rRNA
genes with intervening sequences (Tartof and Dawid,
L976). On the X chromosome, genes carrying inter-
vening sequences appeared to be randomly assorted with
those carrying no such sequence (Pellegrini et al. ,
1977). About 45eo of all the D . rnelanogaster ribo-
somal repeats appear to carry intervening sequences.
Pellegrini et a1. (]-977) observed a short inverted
repeat (100 to 400 base pairs), at the extremities of
the intervening sequence in their "R-loop" studies,
suggesting that these sequences nay be translocatable
elements.
24.
(b) 55 genes
The DNA coding for 55 RNA (5S DNA) can be
isolated from total Xenopus laevis DNA by repeated
cycles of density gradient centrifugation (Brown et
ãL., 1971). Denaturation rnapping of this 55 DNA
indicated that each G + C rich 55 gene was associated
with a longer A + T rich spacer to form a 0.7 kb unit
in a tandernly repeated array of such units (Brown et
a!., 1971). Restriction eîzyme analysis of this
total 55 DNA revealed length heterogeneity in these
A + T rich spacer sequences (Carroll and Brown, L976a).
Repeating units differ from each other by 15 base
pair quanta. (Carrol1 and Brown, 1-976a) and Brownlee
et aI. (1974) have shown that these A + T rich
sequences consisted of tandem repeats of just this
size. Repeat length heterogeneity is therefore due
to variations in the number of these subrepeats
(Carro11 and Brown, I976a).
By cloning of 55 DNA in bacteria and
restriction enzyme analysis, Carroll and Brown (197ób)
showed that adjacent 55 repeats can differ in length.
The l-2l- nucleotide G + C rich 55 gene, however, was
not sufficiently large to account for all the G + C
rich sequences in arLy given 55 repeat unit.
Sequencing studies have revealed the presence of a
"pseudo-gene" structure adjacent to the 5s structural
gene (Jacq et al. , L977). The pseudo-gene appears to
be a perfect copy of 101 residues of the structural
gene. It is presumed that this "pseudo-gene" was
25.
once a structural gene resulting from gene duplication
and produced 55 RNA. Its sequence, however, diverged
sufficiently for the gene to be no longer functional
(Jacq et al., L977). Thus it is thought that the
"pseudo-gene" is a relic of evolution.
(c) Histone senes
The number of histone genes in the genome
of sea urchin varies between individuals, but for those
species studied to date, the histone genes appear to
be reiterated 500 to 1,000 tines (Weinberg 9! aL.,
Lg72; Grunstein et al., L973). Furtherrnore, these
genes appeaï to be tandernly linked as determined by
the analysis of the buoyant density of histone genes
with changing DNA nolecular weight (Kedes and Birn-
stiel , L}T]-). These results suggested that histone
genes r^rere either linked in blocks, each block coding
for one type of histone, or the genes for each of the
histone proteins could be intermingled with each
other. The actual arrangement of histone genes has
been unequivocaLly elucidated by restriction endo-
nuclease cleavage and cloning of the repetitive tan-
dern histone DNA sequences. Restriction analysis of
total sea urchin DNA (q.. purpuratus, L. pictus and
Ps. niliaris revealed the existence of a 6 kb
cluster containing sequences complenentary to all the
histone mRNAs (Kedes et al. , T975; Weinberg et aL',
Ig75; Schaffner et ã1., L976) -
With the
for the different sea
separation of
urchin (Ps.
the rnRNAs coding
his tonesniliaris
26.
(Gross et ãL., L976a), it became possible to obtain a
restriction map of the 6 kb histone repeat. In thismanner, Schaffner et aT. (1976) were able to determine
that the gene sequence within the repeat was H4, HZB,
HS, HZA and Hr. By cloning the repeat unit into À
followed by thernal denaturation analysis of the
amplified DNA, it was shown that about 50% of the
histone repeat consisted of A + T rich DNA which would
not be expected to code for histone proteins
(Schaffner et aI., 1976). It was suggested that this
A + T rich DNA might constitute spacer sequences
between the histone genes of the repeat unit, and this
u¡as confirmed by partial denaturation of arnplified
repeat unit DNA followed by electron microscopic
analysis (Portrnann et ãI. , L976) . This electron
microscopic analysis also indicated that the spacers
between different genes varied in length from 420
base pairs (between H, and HO) to 880 base pairs
(between HO and HrB) (Portmann et aI., L976). Using
restriction enzymes, controlled À exonuclease diges-
tion of cloned DNA and hybridization of purified
histone mRNAs, Gross et aL. (1976b) demonstrated that
the polarity of the DNA fragment carrying the genes
for HrB , H4 and H1
is 5r HrB '> HO * Hl 3t . Since
all highly purified nRNAs hybridize exclusively to
one strand of the cloned DNA, the polarity of the
histone gene cluster is 5' Ht + H2A * HS - HZB * H4 3t.
Sinilar restriction (Cohn et ãL., I976) and electron
microscopic (Wu et ãL., L976) studies on S . purpuratus
have dernonstrated that both the polaríty of the his-
tone repeat and the interdigitation of histone genes
with A + T rich spacers of different lengths are con-
served between these two species. Subsequent
sequencing studies (Schaffner et aT., 1978) have
indicated that histones do not seem to be derived
fron longer precursor proteins, nor is there any evi-
dence for intervening sequences within the coding
regions. The A + T rich spacer segments between
the genes differ from each other, are made up of
relatively sinple nucleotide arrangements, but are
not repetitious, and apparently do not code for addi-
tional large proteins (Schaffner et ã1., 1978).
The histone genes of
ter are repeated 100 times in the
of repeat unit of lengths 4.8 kb
and Hogness, L976; Lifton et al. ,
units code for aLL five histones
the sea urchin situation, HS, HZA
transcribed from one strand, w
transcribed fron the other (Li
The difference between the two
from an additional block of 27
spacer separating Hl from Hr.
three times as frequent as the
D. melanogaster genome (Lifton
)1
Drosophila nel anogas -
genome in two types
and 5.07 kb (Karp
f977). Both repeat
but, in contrast to
and H, genes are
ith H4 and HrB being
fton et ãL., f977).
types of repeat derives
0 base pairs in the
The 5.07 kb unit is
4.8 kb unit in the
et ãL. , L977) .
(d) General conclusions from these gene
arrangement studies
It is interesting to
size heterogeneity in the repeat
note that there is
units of both ribo-
28.
somal and 55 genes. The fact that repeats of these
genes are not identical and the nature of the hetero-
geneity, permit some conclusions to be made about
mechanisms for the maintenance of tandemly linked
genes. Sudden correction mechanisrns such as the
master-slave (Ca11an and Lloyd, 1960) and contraction-
expansion (Brown et ãI., T972) nodels, which require
tandem homogeneity, cannot, therefore, explain these
observations. Sinilarly, the hypothesis that the
evolution of these tandem genes has occurred by gene
duplication followed by nutation and genetic drift is
not supported, since it is unlikely that the variation
produced would be limited to the number of nultiples
of a subrepeat (Carroll and Brown, 1976b). The
regularity of the observed heterogeneity, however,
suggests that some correction mechanism does exist
and the subrepeats of the spacer regions are strongly
irnplicated in the correction process. A mechanisn of
unequal crossing over between mernbers of fanily of
sequences, sirnilar to that proposed by Snith (L973) ,
which leads to the elimination or fixation of variants
while permitting heterogeneity, is a mechanisn for
tandem gene maintenance which is compatible with the
results described for ribosomal and 55 gene repeat
unit length variation. The comparative study of the
spacer sequences of different histone repeat units
frorn the same reiterated group of repeat units is,
therefore, awaited with interest.
Heterogeneous Multiqene Fanilies')
This definition refers to fanilies of genes
29.
which are sinilar but not identical, where the genes of a
given fanily all code for a sinilar funciton. The immuno-
globulin genes and the genes coding for silk noth chorion
proteins are established examples of such multigene families.
(a) Immunoglobulin genes
Most studies have been carried out on
myelorna ce11s producing irnmunoglobulin light chains.
The irnmunoglobulin light chain molecule consists of
two regions, the constant region and the variable
region. It is the variable region of the chain
which plays a part in antibody recognition. Dreyer
and Bennett (1965) first suggested that the constant
and variable regions night be separately encoded in
the genome. This was supported by amino acid se-
quence studies (Hood et a1. , L976) and nucleic acid
hybridîzatíon (Tonegawa et al., L976) studies which
indicated that there \^rere more germ line sequences
coding for the variable region (V genes) than for
the constant region (C genes) for a given type of
irnmunoglobulin chain. The more recent evidence of
somatic re-arrangement of V and C genes confirms this
suggestion (Hozurni and Tonegawa, I976; Rabbitts and
Forster, 1978). Seidman et aT. (1978a) have recently
generated variable region probes against two differ-
ent rc light chain subgroups and with then, identified
two non-overlapping sets of EcoRI fragments of mouse
DNA. Each set consisted of five to ten fragments
and each fragment contained elements of variable region
gene sequences. Since different sets of fragments
50.
carrying V gene elements hrere detected by probes to
different K light chain subgroups, and since 25 to 50
different subgroups have aLready been identified from
amino acid sequence studies, there are probably a
mininurn of 125-150 distinct V genes in the mouse genome.
By cornparing the nucleotide sequences of two V gene
elements and portions of their flanking sequences from
different fragments fron the same subgroup set, exten-
sive regions of homology r^rere observed, with sequence
variations being rnainly linited to those regions cod-
ing for the antigen binding site (Seidrnan et ãL.,
1978a). That this homology extended well beyond the
structural V gene region, was demonstrated by hetero-
duplex analysis using two cloned EcoRI fragments from
the same subgroup set (one 13 kb and the other 5 kb).
A mininum of 2.5 kb from these two fragnents were shown
to be homologous , 0.29 kb of which can be attributed
to the V region element (Seidnan et al., 1978b). No
such extensive stretch of homology I^Ias observed in a
sinilar heteroduplex comparison of two mouse ß globin
genes (Tiemeier et aI., 1978) (where the ß globin
gene is an example of an evolutionarily fixed gene).
Seidnan et a1. (1978b) suggest that this hornology
creates a large target for intragenic recombination.
Such recombination could occur in germ line or somatic
cells expanding and contracting various subgroups
while constantly testing and creating new diversity.
Silk noth chorion proteins(b)
Regier et aI. (1978b) have recently
3L.
reported striking sirnilarities among the amino-
terminal sequences of a set of functionaL1-y relatedstructural proteins from the chorion (eggshell) ofthe silk noth Antheraea polyphenus. The sequence
sinilarities suggest that the proteins are related in
an evolutionary sense, presunably being encoded by
genes which have evolved by gene duplication followed
by diversification via nucleotide substitutions and
re - arrangements .
Hood et aL. (1975) have proposed that
certain groups of functionally related RNAs or pro-
teins could be coded for by families of genes which
are clustered in the genome and evolutionarily related(e.9., immunoglobulin genes (Hood et ãL., 1975)).
Chorion genes are apparently linked and hence would
appear to constitute a second example of an informa-
tional nultigene fanily (Regier et aI.,1978b).
Studies on this nultigene farnily are still in their
early stages. The degree of evolutionary relatedness
of the genes coding for chorion proteins is being
examined by protein sequencing (Regier et ãL., I978a,
b) and sequencing of the mRNAs coding for these pro-
teins. Ultirnately the goal is to understand the
organîzation of chorion genes and their pattern of
expression during developnent.
(c) Keratin genes
From the currently available information
about keratin genes, it seems that they most closely
fit the heterogeneous nultigene farnily classification.
32.
The sinilarities between the chorion protein and kera-
tin systems at the protein level are striking, and
further comparison of these two systems at the gene
organizati-on and control levels are awaited with
interest.
F AIMS OF THE PROJECT
At the time of cornmencement of this work, studies on
the structure of keratin nRNA were in their early stages
and there r^ras preliminary evidence that the keratin genes
in the chick genome were clustered. The initial aims of
the project, therefore, were two-fold.
To study the nature of the 3t end of keratin
rnRNA and the relationship between the unique
and reiterated sequences which appeared to be
present in the messenger.
To investigate nethods for the partial purifi-
cation of genomic keratin sequences and to
examine the possibility of using a density grad-
ient systern, employing RNA:DNA hybrids, to study
the arrangement of tandenly linked genes.
Should this technique prove to be effective, it
r^ras intended to ernploy it in the analysis of
keratin gene organization in the chick genome.
1
2
CHAPTER II
GENERAL }'{ATERIALS AND METHODS
33.
CHAPTER II
GENERAL MATERIALS AND METHODS
A MATERIALS
1. Proteins and Enz mes
Bovine Serum Albunin : Fraction V, Sigrna Chenical Co.,
St. Louis, Missouri.
DNA Polymerase I :. Boehringer Mannheirn, Mannheim,
West GermanY.
Pancreatic Ribonuclease : Type III, Signa.
Proteinase K : E. MerckrDarmstadt, West Germany.
RNA-dependent DNA Polymerase : gift of Dr. J.R.E. Wells,
prepared fron avian rnyeloblastosis virus donated
by Dr. J.W. Beard and the N.I.H. Cancer Program.
Restriction Endonuclease EcoRI : gift of R.B. Saint.
St Nuclease : prepared by the method of Vogt (1973).
2 Radiochemicals
tSHl -deoxycytidine triphosphate (16 - 26C / rnmol) ) : The
Radiochernical Centre, Amersham, Buckinghamshire,
England.
o-[32p] -deoxyadenosine triphosphate (initial specific
activity 9OC/mmo1) : prepared by Dr. R.H. Synons
of this Department.
(initial specific activity 500 rnC/rnl) : The
Radiochemical Centre.
iriu12 5I
34.
3. Chenicals for Sp ecific Procedures
(a) Electrophoresis
Acrylarnide : Merck, twice recrystallized from
cHCl 3
.
Agarose : Signa.
NrN'-rnethylenebisacrylarnide : BDH Chenicals Ltd. ,
Poole, Dorset., England, recrystallized
from CHC1 S.
N,NrNt,Nr -tetramethylethylenediamine : Eastman
Organic Chernicals, Rochester, New York.
Formamide : BDH, deionized as described by
Pinder et al. (1974).
Ethidiurn bronide : Signa.
Toluidine Blue : George T. Gurr, London, England.
(b) Density gradient centrifugatloq
Actinomycin D : gift from Merck, Sharp and Dohme,
Rahway, New Jersey.
Caesium Chloride Optical Density grade : Harshaw
Chenical Co., Cleveland, Ohio.
Caesiurn Sulphate High Purity grade : Kawecki
Berylco Industries, Inc., New York, New
York, Tecrystallized from boiling water
and ethanol.
(c) Complementary DNA synthesis
Deoxyribonucleoside triphosphates : Signa.
Dithiothreitol : Sigma.
Oligodeoxythyrnidylic acid, free acid : P.L Bio -
55.
chenicals Inc., Milwaukee, Wisconsin.
2-Mercaptoethanol : Sigma.
(d) Radioactive counting
NCS Solubilizer : Anersharn/Searle Corpn., Arling-
ton Heights, Il1inois.
P0P0P (1 ,4-bis - (2 , 5 -phenyloxa zoLy1-) -benzene) :
Sigrna.
PPO (2,5-diphenyloxazole) : Sigma.
4 Miscellaneous lt{aterials
Coalfish DNA : Sigrna.
Diethylpyrocarbonate : Signa.
Ficoll : Pharmacia Fine Chenicals, UppsaIa, Sweden.
Nitrocellulose : Sartorius, Göttingen, West Germany.
Phenol : BDH, redistilled under N, and reduced pressure,
stored at -15oC under N2 prior to use.
Polyvinylpyrrolidone : Sigrna.
Sarcosyl : Ciba-Geigy Ltd., Bas1e, SwitzerLand.
Sodium dodecyl sulphate (95"ó pure) : Sigrna.
Sucrose, ultrapure, RNAfase free : Schwarz-Mann,
0rangeburg, New York.
All other chenicals used r{rere of analytical reagent
grade, or of the highest purity available.
5. Preparations of Solutrons
All solutions were prepared in sterile glass
distilled water followed by autoclaving or treatment with
diethylpyrocarbonate to inactivate nucleases. Glassware
nras steriLzed by either autoclaving, incubation at 110oC
36.
overnight, washing with 1 M KOH followed by rinsing with
sterile glass distilled water, or by a conbination of these
procedures. Spatulas, etc., were washed in alkali and
rinsed in sterile water, as described. Pipettes and micro-
pipettes were washed in glass distilled water containing
diethylpyrocarbonate and dried for 16 hours at 110oC.
B GENERAL METHODS
1 Ethanol Preci itation
In all cases samples hrere made 0.1 M with respect
to sodiurn acetate using a 3 M stock solution at pH 5.5.
About 2.5 volumes of redistilled ethanol was added to each
sample, samples were then shaken and stored at -20"C for a
minimum of 2 hours. Precipitates r^Iere then collected by
centrifugation at 15r000 r.p.m. for 15 to 30 min in a Beck-
man J-ZIB centrifuge at 2"C. Ethanol was poured off and
arLy renaining ethanol v/as allowed to drain from the pellet
by inversion of the tube for 20 min. The pe1let was then
redissolved in the appropriate solution.
2. Preparation of Keratin nRNA
dome s ti cus
FertiLized eggs of White Leghorn fowls (Gallus
strain Para. 5 were obtained fron the Parafield
Poultry Research Station of the Department of Agriculture,
Parafield, South Australia. The eggs were stored at 10oC
for no more than 10 days, and incubated at 37"C, 54% hunidity
in a forced draught incubator (Saunders Products Pty. Ltd.,
Adelaide) for 14 days. At L4 days, embryos were removed
from the eggs and washed with Hanks balanced salt solution.
Feathers were plucked into a solution containing 200 nM KCl,
37.
5.5 mM MgCI2,10 mM Tris, pH 7.4, and keratin nRNA isolated
from the nRNP particles produced by EDTA dissociation of
polysomes, as described by Kemp et 41. (1974b). The keratin
nRNA was then purified by repeated cycles of sucrose grad-
ient centrifugation.
3. Preparation of rRNA
The rnRNA preparation procedure involves centri-
fugation of the EDTA treated polysomes on I0-40% sucrose
gradients (Kenp et al., L974b). Large and small ribosomal
subunits r4¡ere isolated from these gradients, the solution
made 5% with respect to sarcosyl and the protein rernoved by
three extractions with an equal volume of a mixture of phenol
and chloroform (1:1). After the final extraction, the RNA
hras precipitated with 2tz, volumes of ethanol at -20"C for a
ninimun of two hours. 28S rRNA fron the large ribosomal
subunit and 18S rRNA fron the small ribosomal subunit were
further purified by sucrose gradient centrifugation (L0-40%
gradients, Beckman SW41 rotorr 37,000 T.p.fl.,16 hours).
Purified RNA was ethanol precipitated twice and stored as
an ethanol precipitate until required.
4. Preparation of Chick DNA
Chick blood was obtained by cardiac puncture
(for large volumes) and from a wing vein (for small volurnes).
Blood was collected into tubes containing 0.15 M NaCl, 5 mM
KCl, 2 nM MgCIZ (NKM) made I% w/v with respect to heparin.
The blood was centrifuged at 2,000 r.p.rn. for 5 min at 4"C
and plasma and white celIs removed. The erythrocytes were
then washed a further two times with NKPI. Approximately 5
58.
volumes of 2 nM lvlgCl Z \ras added to the packed erythrocytes
and stirred at OoC for 10 min to bring about lysis of the
cell membrane. Nuclei were spun down at 4r000 r.p.m. for10 min at 4oC and the lysis procedure repeated.
High molecular r^Ieight DNA was extracted from
the erythrocyte nuclei by a nodification of the procedure
described by Gross-Bellard et al. (L973). A portion of
the nuclear pellet was lysed in a solution containing 10 rnM
Tris-HCl pH 7.5,10 mM EDTA, 10 mM NaC1, 0.5% SDS and Pro-
teinase K at a concentration of 200 Ug/ml. The solution
was incubated at 37"C overnight then extracted once with
water saturated phenol and twice with a mixture of phenol
and chloroform (1:1) . The aqueous phase was dialysed
against 3 changes, each of 2 litres of 10 nM Tris-HCl, pH
7.5, 1 nM EDTA (TE). Pancreatic ribonuclease A, previously
heated to 80oC for 10 min, was added to a concentration of
10 vg/nI and the solution incubated at 37"C for a ninimr.un of 2 hours.
The DNA-containing solution was then made up to 10 mM Tris-HCl pH 7.5, 10 mM EDTA, 10 rnM NaCl, 0.5% SDS and Proteinase
K added to a final concentration of 100 ug/nl. The solution
hras incubated for a minimurn of 2 hours at 37"C then extracted
5 times with a mixture of phenol and chloroforn (1:1) . The
DNA solution r^ras dialysed against 3 changes, each of 2 litres,
of TE and then stored at 4"C.
To obtain DNA of lower molecular weight, DNA
'hras extracted from chick erythrocyte nuclei by a nodification
of the method of Marmur (Marushige et aL., 1968).
59.
5 Preparation of Hydroxylapatite (HAP)
500 ml Each of 0.5 M CaCl, and 0.5 M NarHpOO
\4rere slowly nixed with stirring over a period of about z
hours. CaPO4 crystals were allowed to settle and the
supernatant removed. crystals r rere washed 4 times with750 nl of glass distilled water. After the final wash,
750 nl of hot water was added to the crystals along with25 ml of a freshly prepared 40% (w/w) solution of NaOH.
The nixture was boiled gently with stirring for t hour then
removed frorn the heat source and the crystals allowed tosettle for 5 rnin. The turbid supernatant was removed and
the crystals washed 4 times with water. After the fourthwash, 1 litre of 0.01 M phosphate buffer, pH 7.0 was added
to the crystals. The mixture r^¡as heated with stirringuntil boiling began. The supernatant was removed and the
procedure repeated as before except that the boiling was
allowed to continue for 5 nin. The supernatant was again
removed and the procedure repeated with boiling for 15 nin.The supernatant was removed and the HAp crystals stored in10 mM phosphate buffer, pH 7.0, at 4"C.
6. Preparation of cDNA
(a) cDNA to keratin nRNA
DNA conplementary to keratin nRNA was
synthesized in a system (25 ul) containing 50 nM Tris-HCl, pH 8.5, 8 nM dithiothreitol, 8 nM lr,IgCl2, 0.66 m}I
each of the three unlabelled deoxynucleoside triphos-phates, 60 uM 1abel1ed deoxynucleoside triphosphate
(32p dATp or 3H dCTp), 100 uglml actinomycin D, 0.5-1
40.
ug rnRNA, 0.1 ug oligo dT and AMV-DNA polymerase.
Syntheses'hrere carried out at 37"C for t hour in the
early part of these studies, but subsequently it was
found that in such preparations, anything up to 15%
of the total cDNA could be complementary to rRNA.
This contamination could be reduced to insignificant
leve1s by allowing the synthesis at 37oC to proceed
for 15 nin. After incubation, the RNA tenplate was
removed by alkaline hydrolysis with 0.5 M NaOH for 1
to 2 hours at 37"C. The cDNA was separated from
unincorporated triphosphates by chromatography on
Sephadex G-50, in 0.2 M NaCl, 1 nM EDTA, 10 nM Tris-
HCl, pH 7.5 followed by ethanol precipitation.
(b) cDNA to rRNA
Synthesis of cDNA to rRNA also employed
the reverse transcriptase of avian rnyeloblastosis
virus, but used random oligonucleotides of salmon
sperm DNA, prepared as described by Taylor et al.
(L976), to prirne the synthesis at non-specific points
along the 18S and 28S rRNAs.
Using saturating amounts of enzyme, cDNA
was synthes ized in a ZS ul reaction volume containing
100 mM KCl, 3 mM MgCl2,8 rn}4 dithiothreitol, 0.67 mM
each of dGTP, dTTP and dATP and 60 uM 3H dCTP, 50 rnM
Tris -HCl, pH 8. 5, 100 ug/rnl actinornycin D, I ug of
RNA, 50 ug of primer DNA and 1 ul (2.5 units) of
eîzyme. Reaction mixes l4¡ere incubated at 37"C for
60 min and cDNA isolated as above.
4L.
7 Radioiodination of RNA
1 ul2vr1 ul1ul
Radioiodinations were performed by a modification
of the nethod of Prensky (1976). Radioiodinations r,rrere
carried out in a 10 U1 volurne using I to 2 Vg of RNA. A
typical reaction mix is set out below with the important
final concentrations in parenthesis.
1 rng/rnl herparin
RNA
*Acidification rnix
Na l25r
0.1 M T1C1S in 1 M
H2
HNos, diluted r/zo
inlMsodiunace-tate pH 4.7
0
1ul4ul
*Acidification mix = 0.75 M HNO
acetate pH 4.7 (Prensky , L976) .
(0.1 mg/ml)
(L-2 ug)
(tI-l - 3 x 10 s - 1x4oo - 5oo uC of 125I)
(Sodiurn acetate 0.1 M
[T1** ] = 5 x 10-4 M)
s, 1o-4 M Kr , 0.4 M sodiurn
-4,10 M
Heparin was used as a ribonuclease inhibitor.
The acidification mix was used to neutralize the Ntl25t
since it is delivered in a solution of pH between 9 and 11.
This mix also contains KI to allow the concentration of I-
to be raised to within the optimum range (Prensky, 1976).
The reaction mix was incubated at óOoC for 10
min then chilled on ice. 1 ul Of 2-mercaptoethanol was
added and the iodinated RNA was separated frorn the unincor-
porated L25I by chromatography on Sephadex G-50 in 0.2 M
NaC1, 1 mM EDTA, 10 mM Tris-HCl, pH 7.5. Fractions con-
taining iodinated RNA were pooled and ethanol precipitated.
The radioiodinated
ofbetween 2xir07
42.
products generally had specific activities
and8xT07c.p.m./ug.
B. Deterrnination of Radioactivity
For liquid hybridization and reassociation
studies, samples hiere precipitated with I0% trichloroacetic
acid, in the presence of bovine serum albumin or coalfish
DNA as carrier, ãt OoC. After 50-60 min, the precipitates
hrere collected on glass f ibre f ilters (Whatrnan, GF/A) .
Filters were washed twice with 5 nl of 5% TCA then exten-
sively with ether, placed into 2.5 m1 glass vials and dried
at 110"C for a ninimum of 30 min. Toluene-based scintilla-
tion fluid (3.5 g PPO, 0.35 g POPOP per litre of toluene)
hras then added, and the samples counted in a Packard liquid
scintillation spectrometer. Nitrocellulose filters from
filter hybridizations, after washing and drying, were counted
in a simiLar way.
In cases where sma1l volurnes (up to 10 ul) of
aqueous samples were to be counted, these l^Iere dissolved in
2 mL of "toluene-triton" scintillation fluid (5 volumes of
toluene scintillation fluid plus 5 volumes of triton X - 100).
3H 1abel1ed radioactive products in ge1 slices
were detected after solubilization of each slice with 0.2 nl
NCS solubilizer plus 0.025 ml of 8 M NH40H, to which was
added 2 mL of scintillation fluid. Gradient fractions and
gel s lices containing 32p 1abel1ed products r^rere determined
by Cerenkov counting.
Sarnples were counted for as long as practicable
in order to ninimize counting errors.
43.
I Disestion of À P Restriction
Endonuclea'se
EcoR, digestions of À phage DNA were performed
in a 16 pl incubation mix containing I ug of I DNA, 100 nM
NaCl , 10 nM MgC12, 10 nM 2-nercaptoethanol , 10 rnM Tris-HCl,
pH 7 .5 and 2 units of EcoRI restriction endonuclease.
(One unit of enzyme conpletely digests 1 ug of )' DNA in 60
min.) Incubations were carried out at 37"C for 60 rnin and
the reaction was stopped by standing on ice.
CHAPTER III
STRUCTURE OF KERATIN NRNA
44.
CHAPTER III
STRUCTURE OF KERATIN NRNA
A. INTRODUCTION
The work described in this chapter deals with studies
on the 3t end 150 nucleotides of keratin nRNA and their
relationship to the rest of the nolecule. This work was
begun by D.J. Kernp, formerly of this laboratory, and a com-
plete appreciation of the study will require the presentation
of a lirnited amount of data obtained by hirn and published
j ointly (Kemp, Lockett and Rogers, in preparation) ' Data
obtained by Kenp will be indicated in the text '
Feather keratin has distinctive properties relevant
to studies on the organization of different classes of se-
quence. Feather keratin consists of a large number of
homologous polypeptide chains, all containing about 100 amino
acids but differing in primary structure by rnultiple amino
acid substitutions (o'Donne11, L973b; O'Donnell and Inglis,
Lgl4; Kernp et al., 1975). Highly purified keratin nRNA
(Partington et àL., Ig73; Kenp et al. , L974b; Kenp et a1 ',
I974c) was about 0.8 kb long, indicating the presence of
0.5 kb in addition to the keratin coding sequence of 0'3 kb'
Keratin nRNA acted as a ternplate for the synthesis of com-
plementary DNA (cDNA) by the RNA dependent DNA polymerase
of avian rnyeloblastosis virus (Alw) (Kemp et al., 1975;
Kernp, 1975). Hybridization of keratin cDNA to keratin
nRNA indicated that the sequence complexity of keratin nRNA
was 25 to 35 times that expected for a single molecular
species (Kernp, 1975). The apparent rate of hybridization
45.
r^ras about ten tines greater when assayed on HAp than when
assayed by st nuclease and the hybrids increased in thermal
stability with increasing Rot. These results were explained
by the hypothesis that each of the honologous keratin rnRNA
species contained a common (but non-identical) sequence
(reiterated sequence) covalently attached to a sequence so
different in base sequence from that of ar:y other keratinnRNA species (unique sequence) that it appeared unique inthe chick genome, under the experimental criteria used.
It was estinated that there r^rere about 100 to 240 copies ofthe reiterated sequence in the chick genome and suggested
that this sequence is the keratin coding sequence (Kemp,
1e7s).
It was possible to obtain shorter keratin cDNA molecules
which hrere representative of the 3t end of keratin nRNA, by
reverse transcribing the nRNA with E. coli DNA polymerase I(PolI). Using this short cDNA as probe, it was possible to
determine the nature of the 3t end sequences and whether
they were representative of the unique or reiterated class
of sequences.
Initial studies involved the comparison of the hybrid-
ization kinetics for PolI- and AMV-cDNAs to keratin mRNA
using the equivalent hybridizations in the rabbit globin
system for kinetic standards. While the observed kineticsof hybridization for AMV-cDNA to keratin nRNA r4rere fasterwhen assayed on HAP than when assayed by nuclease St (Kenp,
1975), no such change in rate was observed when keratinPolI-cDNA was used as probe (Kernp et al., in preparation).
Denaturation studies showed that, while both keratin AMV-cDNA
46.
hybrids and PolI-cDNA hybrids changed in thermal stability
as a function of Rot, the change observed for the PolI-cDNA
hybrids was nowhere near as marked as that for the AMV-cDNA
hybrids (Kernp et al ., in preparation). In addition, unlike
the results with AMV-cDNA, the PolI- study showed no signi-
ficant transfer of hybrids from low thernal stability to
high thernal stability (Kenp et ã1., in preparation). The
T- of the PolI - cDNA hybrids r^ras about 83oC after completem
hybridization compared to 90oC for AMV-cDNA hybrids. At
least a part of this T, reduction was probably caused by the
shortness of the hybrid, since globin PolI-cDNA hybrids
showed a 4"C lower T, than globin AI4V-cDNA hybrids (Kernp
et al. , in preparation) .
These results suggested that the PolI-cDNA had differ-
ent characteristics from the AMV-cDNA which inferred that
different regions of the nRNA molecule night have been sub-
j ect to different divergence pressures. This regional
variation r^¡as , therefore, studied in more detail .
B METHODS
1. Preparation of rnRNA, rRNA, cDNAs and DNA
Keratin nRNA was prepared fron EDTA treated
polysomes as described by Kemp et al. (1974b). Ribosomal
RNA was prepared as described in Chapter II.B.3. AIvIV-cDNAs
to keratin nRNA and rRNA were prepared as described in Chap-
ter II.B.6 (a) and (b) respectively. PolI-cDNAs hlere syn-
thes ized in a system (50 ul) containing glycine buffer , (67
mM, pH g.2) KCl (50 mM), dATP, dTTP, dGTP (50 uM each), tr-
dCTP (20 uM) (Sp. Act. 26.2 mC/prnole) 2-mercaptoethanol (10
47.
mM), Ivhclz (0.5 mM), dTto Q.5 uglnl), rnRNA (0.6 ug), act-
inonycin D (100 ve/nL) and E. coli DNA polymerase I (1 ug)
incubated for 90 nin at 37oC and purified as for AMV-cDNA
(personal communication of 0. Bernard). Chick erythrocyte
chromosomal DNA was prepared essentially as described by
Marushige, Brutlag and Bonner (1968) and sonicated to aî
average length of about 500 bases as deterrnined by electron
microscopy. Chick DNA was labe11ed by "nick translation"
with E. coli DNA polymerase I in the presence of DNAse I as
described by Schachat and Hogness (1973), using 3H-dCtp (Sp.
Act. 26 C/nnole) as the only Iabelled deoxynucleoside tri-phosphate. After labelling, the DNA was purified as for
cDNA. PolI-cDNA was elongated as described by Crawford
and Wells (1976).
2. RNA-cDNA Hybridizations
(a) Hybridization kinetics
18S rRNA, 285 rRNA and keratin nRNA r^rere
used in the hybridi-zation kinetic analysis. Hybrid-
izations were perforrned at 60oC for 4 hours in 0.18 M
NaCl, 0.001 14 EDTA, 0.01 M Tris-HCl, 0.05% SDS, pH 7.0.
The reaction mixture s (25 ul) contained 2 ,00 0 c.p.rn.
of cDNA and varying amounts of RNA. For St nuclease
assays, 25 ul reaction mixes l\Iere <liluted into 1.0 ml
of 0.03 M Na-acetate, 0.05 M NaCl, 0.001 M ZnSOO, 5%
glycerol, pH 4.6 (1ow salt St assay buffer) containing
LZ ug of sonicated, denatured coalfish DNA. An
aliquot of 0.5 ml was rernoved from each sample and
identical samples were incubated with or without St
nuclease (4 units, Vogt (1973)) at 45oC for 30 min,
48.
TCA precipitated (Chapter II.B.8) and counted.
(b) Estination of ribosomal contaminants in
cDNA preparations
Hybridizations to determine the percentage
of PolI- and AMV-cDNAs complementary to rRNA were
carried out by nixing 1.3 ug each of 18S and 28S rRNA
with 4,000 c.p.rn. of the appropriate 3H labelled cDNA
in a final volume of 20 u1 of the same hybridizatîon
buffer as above. Sanples r^¡ere sealed in sterile 50
ul microcapillaries, placed in boiling water for 2
min, then incubated at 60"C for 18 hours giving a Rot
value of about L4 mol.s.1-1 fot each ribosomal species.
Sinilar hybridizations r^rere performed using 0.5 Ug of
keratin nRNA giving a Rot value of about 5 no1.s.1-1. llnder
these conditions all hybridizations should have gone
to completion. The hybrids r^rere then assayed by St
nuclease in the above low salt St nuclease digestion
buffer as described by Kernp (1975), and the percentage
of the counts in hybrid form estinated.
(c) Competition hybridizations
Conpetition hybridizations were carried
out in the same buffer as described above. 25 ulAliquots of a hybridization solution containing a
molar ratio of cold AMV-cDNA:keratin mRNA:3H 1abe1led
PolI -cDNA of 100: 40 :1, or of nRNA: PolI -cDNA of 40 : 1,
hrere sealed in sterile 50 ul microcapillaries, placed
in boiling water for 1 nin and then transferred to a
60oC waterbath. Aliquots were removed at different
49.
tines, snap chilled in ice r¡¡ater and stored ftozen at
-20"C. When the last sample had been removed,
earlier samples hlere thawed and assayed by nuclease
St as previously described (Kemp, 1975).
3. cDNA-DNA Reassociations and Thermal Stability
Analyses
(a) Reassociation
Reassociation of keratin AMV- , PolI - and
elongated PolI-cDNAs with sonicated chick erythrocyte
DM,10-15 mg/ml,was performed in the same buffer as described in
Chapter III.B.2. (a), using a ratio of 2-3,000 c.p.m.
cDNA to 1 mg of DNA. Reaction mixtures were heated
to 100oC for 5 min to denature the DNA prior to incu-
bation at 60oC to the appropriate Coa values. For
St nuclease assays, 50 ul time samples L\Iere diluted
into 1 nl of 0.05 M Na-acetate, 0.3 M NaCl, 0.001 M
ZnSOO , Seo glycerol, PH 4.6 (high salt St assay buffer) ,
the sarnples divided into tI^Io and incubated at 37"C
with or without nucleas" 51 (150 units). For HAP
assays, 50 pl aliquots l^Iere diluted into 2 mL of 0.I2
M phosphate buffer (0.18 Ir't Na+) . Single and double-
stranded DNA were then separated on HAP at 60oC as
described by Britten and Kohne (1968). After St
nuclease digestion or HAP fractionation, aI1- samples
for aîy one curve were made up to the same final con-
centration of DNA before TCA precipitation to avoid
eïrors resulting from differential quenching.
50.
(b) Thernal stability
For the determination of the thernal
stabilities of cDNA-DNA duplexes, 50 ul aliquots
taken at different values of Cot were diluted into 2
ml of 0.LZ M phosphate buffer and loaded on to HAP at
60oC. Duplexes which could not be eluted at 6OoC
r^/ere melted by raising the tenperature in 5oC incre-
ments. Two 2.5 ml washes (0.12 M phosphate buffer)were collected at each temperature and TCA precipitated.
4 Pre'paration of Keratin AMV- and PolI-cDNA
Duplexes for Density Gradient Centrifugation
AMV- or PolI-cDNA duplexes r^¡ere prepared by
incubation to a Cot of 2 x 104 mol.s.1-1 ",
above. Sanples
(100 ul) containing 1 ng DNA and 5,000 c.p.m. cDNA \^rere
added to 4.0 m1 high salt nuclease St assay buffer and incu-
bated at 37"C for 15 nin with 100 units of nuclease S1
The reaction was stopped by addition of 0.ZS nl of 0.4 M
+Na' phosphate, pH 7.0 and the nixture was dialysed against
3 changes of buffer (10 mM Tris-HCl, pH 8.4, 1 mM EDTA, 10
mM NaCl). Chick DNA (10 kb) rtras added to a total of 1 ng
DNA and buffer and solid Cs,Cl were added to a volume of Lz
nl and a density of L.6 gn/cc. The mixtures r^rere chilled
on ice and 500 ug actinornycin D was added. The refractive
index was adjusted to 1.3903 at 20"C. The mixtures were
centrifuged in a Beckman Ti50 rotor at 321000 r.p.m. for
60 hours at 20"C. The gradients were fractionated by up-
ward displacement and 0.5 rnl fractions were collected. The
AZOO of each fraction was determined and the amount of DNA
in each fraction was made constant by addition of sonicated
51.
calf thyrnus DNA. Actinomycin D was extracted with CsCl-
saturated isopropanol. Water (1.5 rnl) and 20% TCA contain-
ing I% sodium pyrophosphate (2.0 ml) were added to each
fraction. The precipitates were collected on Whatman GF/A
filters, rinsed twice with 5% TCA - L% sodiun pyrophosphate,
twice with ether, dried and counted in toluene-based scin-
tillation fluid.
C RESULTS
1 nRNA Purity
The work described in this chapter depends on
the PolI- and AMV-cDNAs being representative of particular
regions of the keratin rnRNA molecule. It was, therefore,
necessary to prove either that the messenger preparations
were completely pure or that aîy contaminating sequences
were not copied by the enzymes. It has been published pre-
viously (Partington et al., L973; Kernp et al., L974b; Kemp
et ãI., I974c) that keratin nRNA preparations rlrlere greater
than 95% pure as assayed by formanide ge1 electrophoresis
and in vitro translation. While the latter piece of infor-
mation suggests the absence of contaminating messenger
species, these methods of analysis would not have detected
rRNA breakdown products of the same size as the 1RNA (0.8
kb). Contaminating ribosomal sequences could be detected,
however, by liquid hybridizatíon using cDNA made to rRNA by
the random prirning nethod of Taylor et al. (L976) (see Chap-
ter II.B.6.(b)). Complementary DNA was nade to purified
18S and 28S rRNA (see Chapter II.B.5.) and the kinetics of
hybr idization of these cDNAs to keratin rnRNA rlilere compared
to those for the homologous reactions (i.e., 18S rRNA with
52.
cDNA made to 18S rRNA etc.). The results are shown in
Figure 3.1. By dividing the *otU for the homologous hybrid-
ization by the *oa4 for the heterologous reaction, the pro-
portion of the keratin nRNA preparation consisting of 18S or
285 rRNA breakdown products could be deternined.
Frorn Figure 5.1(a) the Ooa% for the hybridiza-
tion of cDNA, made to 28S rRNA, to 28S rRNA was L.2 x l.0-2
mol.s.1-1 while the Rotþ, for the hybridization of the same
cDNA to keratin nRNA was 5.5 x 10-1 mol.s.1-1. The propoï-
tion of the keratin rnRNA prepatation which could be attri-L.2 x 10
_)buted to 285 rRNA breakdown products r,rlas therefore -15.5 x 100.022 or 2.2% of the preparation. From Figure 5.1(b) the
R^tr. for the hybridization of cDNA made to 18S rRNA to 18So4rRNA was 2.6 x 10- 5 mol. s.1 -1. The Rot
tion of this cDNA to keratin nRNA was 9
Thus the proportion of the keratin nRNA
able to 1BS rRNA breakdown products I^Ias
U for the hybri-diza-
x 10-5 mol.s.1-1.
preparation attribut-2.6 x 10 t
-
- ¡.2999 x 10-J
or 28.geo of the keratin nRNA preparation.
It was apparent from these results that approx-
irnately 30% of this particular nRNA preparation was derived
from ribosomal breakdown material with most of the contam-
inant coming from 18S rRNA. Although the actual proportion
of any given nRNA preparation that could be attributed to
ribosomal breakdown products would be expected to vaTy from
preparation to pïeparation, 7t least 20% of the RNA in aîy
particular preparation could be expected to be of ribosomal
origin.
FI GURE 3. ]. .
THE DEGREE OF C ONTAMINATION OF KERATIN NRNA PREPARATTONS
WITH 18S AND 28S rRNA BREAKDOWN PRODUCTS
cDNA made ro 18S or 28S rRNA (Chapter II.8.6. (b)) was
hybridized to keratin rnRNA (Chapter III.B.2. (a)). The
rates of these heterologous hybridizations as assayed by St
nuclease, were conpared with the rates of the homologous
ribosomal hybridizations (e.g., 28S RNA to cDNA made to 28S
rRNA) assayed in the Same Inlay. The reactions l4lere carried
out with a vast excess of RNA over cDNA. Sanples were
denatured at 100oC and hybridizations carried out at 60oC.
(a) cDNA to 28S rRNA hYbridized to:
o 28S rRNA; o keratin nRNA.
(b) cDNA to 18S rRNA hYbridized to:
O 18S rRNA; o keratin nRNA.
0
00
0
I
g
.9
fil
.I3l¡
I{o
20
¡10
60
80
20
40
60
80
100 t I04 -3 -2
Log Rot-5
ao
oo
o
a
a
o
o
o
b
O
a
E
o
ooo
53.
The Nature and Specificity of AMV- and poll-
cDNAs
The synthesis of AI,{V- and poll-cDNAs both employ
oligo-dT as a primer (see chapter rII.B.1 for polr-cDNA
synthesis conditions). since keratin rnRNA has a 60 base
Poly(A) tract at its 3t end (Morris and Rogers, in press)
while rRNA has no such structure, the AMV and polI enzymes
would be expected to copy only keratin nRNA, provided cDNA
synthesis could be shown to be completely dependent on the
presence of the primer. The primer dependence of polÏ-
and AMV-cDNA synthesis, using keratin nRNA as a template,
hras therefore examined. In addition, the efficiency withwhich these enzymes copied rRNA, in the presence and absence
of oligo-dT, hlas also studied since rRNA breakdown products
constituted a significant proportion of the keratin nRNA
preparation. The results are shown in Table S.1.
The PolI section of Table 5.1 shows that copy-
ing of keratin nRNA was almost completely dependent on the
presence of oligo-dT primer. It also shows that poll
copied rRNA quite readily in the presence or absence ofoligo-dT primer. This latter result suggests that the 7.7%
background synthesis observed using keratin nRNA as ternplate
in the absence of oligo-dT, might be attributed to copying
of the contaminating ribosomal sequences. Presurnably, how-
ever, 92.3% of the cDNA copied from keratin rnRNA in the pre-
sence of oligo- dT represents keratin sequences and so the
1ow contamination with ribosomal cDNA would not significant-Iy affect aîy of the conclusions made where PolI-cDNA has
been used as a probe.
2
TABLE 3.7
DEPENDENCE OF POLI AND AMV COPYING OF KERATIN mRNA
ON THE PRESENCE OF OLIGO dT PRIMER AND THE RELATIVE
EFFICIENCIES OF THESE ENZYMES FOR COPYING TRNA
Experirnental details
C.p.n. incorporated intoTCA precipitable naterialas a percentage of those
incorporated in the enzyme+ primer + keratin nRNA
synthes isPolI +
Po1 IPolI +
Po 1I
dTdTdTdT
oligoo1 igooligooligo
100 %
3.s%2 .0%xL.7%*
dTdTdTdT
oligooligooligooligo
+ keratin nRNA+ keratin nRNA+ rRNA+ rRNA
100 %
'7 10,*T. T'O
58.8%52.9%*
AMVAMVAMVAMV
+ keratin mRNA+ keratin nRNA+ rRNA+ rRNA
PolI-cDNA synthesis was as described in Chapter III.B.1. except that mixes \^rere made up + oligo dT in 10 Ul
reaction volumes. 0.5 ug Of purified nRNA or 0.25 ug each
of 18S and 28S rRNA was used as template in the reaction
mix in the presence or absence of oligo dT primer.
AMV-cDNA syntheses were as described in Chapter II.B.6.(a) except that mixes hrere made up + oligo dT in 10 ulreaction volumes. Ternplates r^rere as for PolI.
*Denotes that the result is the average of two indepen-
dent experiments.
+
+
54.
The AMV section of Table 3.I shows that AMV-cDNA
synthesis hras alrnost conpletely dependent on the presence
of oligo-dT prirner and that no significant copying of rRNA
occurred, regardless of the presence or absence of oligo-dT
primer.
Keratin PolI- and AMV-cDNAs r,'rere also hybridized
with excess rRNA to provide an independent estimate of the
proportion of the cDNAs complementary to rRNA. Table 3.2
shows that only 3% of the PolI-cDNA and 2.L5% of the AIUV-
cDNA hybridized to rRNA. In contrast 90.9% and 75.2% of
PolI- and AMV-cDNAs respectively hybridized to keratin nRNA.
Although these latter values might have been expected to be
100%, in practice keratin A¡[V-cDNA:nRNA and PolI-cDNA:nRNA
hybridization reactions have never been observed to exceed
about 80% and 90% hybridization respectively when assayed
by St nuclease. Rabbit globin hybridízations, however,
always went to 100%.
Thus reverse transcription of keratin nRNA by
both PolI and Alt4V reverse transcriptase r4ras cornpletely de-
pendent on the presence of oligo-dT in the synthesis rnix.
In addition, the proportion of the cDNA complernentary to
rRNA made by either enzyme was insignificant.
3. Molecular Weights of Keratin AIUV-cDNA and PolI-
cDNA
Keratin cDNA was prepared using reverse trans-
criptase (Alv[V-cDNA) and also using E. coli DNA polymerase I
(PolI-cDNA). The molecular weights of the keratin AMV-cDNA
and PolI-cDNA were estimated by polyacrylanide ge1 electro-
TABLE 3.2
DETECTION OF REVERSE TRANSCRIPTS OF TRNA IN KERATIN
'POLI- AND AIW-c.DNAs BY LIQUID HYBRIDIZATION
Hybridizat,ion % Counts resistant to nuclease S
PolI-cDNA:PolI-cDNA:
AMV-cDNA :
AMV- cDNA :
rRNAkeratin nRNA
rRNAkeratin nRNA
2.LS%r375.2%
3%*90.9%
Keratin AMV-cDNA or PolI-cDNA was hybridized to rRNA
or keratin rnRNA as described in Chapter III.B.Z.(b) and
assayed using 51 nuclease in low salt assay buffer (Chapter
III. B .2. (a)) . The resistances of PolI- and AMV-cDNAs to 51
nucl-ease were 5% and 7eo respectively and the figures in the
table have been normali zed to account for this as in Kernp
(1e7s).
*Denotes that the result is the average of two indepen-
dent experiments.
55.
phoresis in the presence of fornanide (Pinder, Staynov and
Gratzer, L974) using RNA markers (Figure 3.2). Both cDNA
preparations were heterogeneous with regard to size. How-
ever, the main peak of keratin AMV-cDNA was of apparent MW
Etu r.7 x 10" (about 0.55 kb) while that of keratin polr-cDNA
r^ras of apparent MW n, 4. 5 x 104 (about 0.15 kb) . Since
transcription of nRNA into cDNA in the presence of oligo-dThas been shown to initiate at the poly(A) region of nRNA
(Milstein et ãI., L974; Proudfoot and Brownlee, 1974 Craw-
ford et ãL., 1977) and since PolI-cDNAs are shorter than
the respective AMV-cDNAs (Milstein et aI., I974; Proudfoot
and Brownlee, 1974; l4odak et al., 1973; Gulati et a1., L9T4;
Rabbitts, I974; Crawford et aL., L977), keratin PolI- and
AMV-cDNAs were expected to be different length copies ofkeratin nRNA, both being initiated at the 3t end of the nRNA.
The poly(dT) segments of chick globin PolI- and AMV-cDNAs,
prepared under conditions identical to those used here, have
been shown to average L8-20 bases in length (Crawford et gL.,
1977) so the keratin PolI- and AMV-cDNAs are presurnably com-
plementary to regions of keratin nRNA extending about 0.15
and 0.53 kb from the poly(A) tract respectively.
4 Conpetition Hybridization Between PolI- and AMV-
cDNA
To be certain that the AMV reverse transcriptase
and DNA polymerase I were copying the same set of nRNA mole-
cules, the AMV- and PolI-cDNAs hrere tested for the presence
of common sequences by studying the competition between the
two cDNAs for sequences in the nRNA during hybridization.Figure 3.3 shows the hybridization kj-netics of 3H-1rb"lled
FI GURE 3 .2 .
POLYACRYLAMIDE GEL ELECTROPHORESIS OF KERATIN AMV-CDNA
AND POLI-cDNA IN THE PRESENCE OF FORMAMIDE
Aliquots of AMV-cDNA and PolI-cDNA were co-electro-
phoresed with the marker RNAs on 4% polyacrylarnide gels in
fornanide containing 0.02 M Barbital, pH 9.0, by the proce-
dure of Pinder et al. (1974) . After electrophoresis for
t hour at 5 nA/gel, the gels I4lere stained with 0.05% Tolui-
dine blue, photographed, and cut into 1 nm slices. The
slices r{rere counted in Toluene-NCS scintillation fluid.
The c.p.m. scale for PolI-cDNA was half that shown.
o
of markers
are shown.
AMV-cDNA o PolI-cDNA. The positions
(rabbit reticulocyte 285, 185, 55 and 45 RNAs)
106
15
+.
.9o3s
.E
ogoE
10
Na'oil-xEct(,
.Ë
.ì+,oct.9Eßt
É,
510
5
4020 30D istance (mm )
10
5S4S
åss' \ 18S\+
56.
PolI-cDNA to nRNA in the presence and absence of cold com-
peting AMV-cDNA. The molar ratio of cold AMV-cDNA:nRNA:
labelled PolI-cDNA was 100:40:1. As described by Gambino
et al. (L974), the percentage of a radioactive probe induplex forn at the end of a DNA reannealing reaction is de-
pendent on the ratio of the reacting species. Consider
the DNA strand with the same sequence as the cDNA as the +
strand and the strand with the same sequence as the RNA,
the - strand. In a reannealing reaction there would always
be equal quantities of the cold reacting + and - species
since DNA is double-stranded. If only a trace amount of
cDNA probe was used relative to the amount of the cold +
strand in the reaction, the extent of the reassociation, as
measured by the radioactive probe, would be almost 100% at
conpletion. If, however, the cDNA probe was present in the
same amount as the cold + strand, there would be 2 times as
many + strands as - strands. Thus, the chance of incorpor-
ating a labelled cDNA into a duplex is one in two and so the
observed extent of the reaction will be 50%. If the amount
of cDNA was further increased, then the excess of + strand
over - strand would also increase and so a decreasing pro-
portion of the input radioactively 1abe11ed probe would be
incorporated into duplex form. In fact, under these con-
ditions of + strand excess, the additional cDNA rather than
the cold sequences alone, drives the reannealing reaction.
In an analogous manner, when a trace amount of labelled cDNA
is used in an RNA driven DNA:RNA hybridization reaction, the
observed extent of hybridization would be expected to be
L00%. This should remain true for ratios of cDNA:RNA up
to 1:1. When the amount of cDNA exceeds the amount of
FIGURE 3.3
3HYBRIDIZATION KINETICS OF H LABELLED POLI-CDNA TO KERATIN
rnRNA IN TH E PRESENCE AND ABSENCE OF COLD COMPETING AMV-CDNA
Hybridi zation solutions containing molar ratios of
cold AMV-cDNA : nRNA : PolI-cDNA of 100:40:1 and nRNA , 3H
labelled PolI-cDNA of 40:1 were divided into ZS Ul aliquots
and hybridizations performed as described in Chapter III.B.
2 . (c). Hybridizations l^rere assayed by nuclease Sl.
Curves \4rere normalîzed to Temove background (Kenp, 1975).
The normalization faclor is shown in parenthesis for each
curve.
T Hybridization in the presence of cold AMV
competitor (5%).
Hybridi zation in the absence of cold AMV con-
petitor (5%).
0
.9
ßtN
=L-ct
20
40
ì 60I
oo
80
100-4 -3 -2 1 0
Log Rot (mol.s.l-r)
57.
driving NA, the proportion of probe which can be in hybrid
form is dependent on the ratio of cDNA to RNA. If twice as
much cDNA as RNA is used in the hybridization, only half of
the probe can be in hybrid form at completion. In the
experirnent shown in Figure 3.3, 2.5 times as much cold AMV-
cDNA was present in the hybridization as rnRNA. Thus itwould be expected that only 40% of the AMV-cDNA could be inhybrid forn at completion. Using a trace amount of PolI-
cDNA as probe, the extent of PolI-cDNA hybridization at com-
pletion would also be expected to be 40%, if the AMV- and
PolI-cDNAs were copies of the same nRNA population. (The
same result would be expected if a trace amount of labelled
AMV-cDNA was used instead of PolI-cDNA.) If AMV- and PolI-
cDNAs represented copies of different nRNA populations, how-
ever, the extent of PolI-cDNA hybridizati-on at completion
would be expected to be L00%, regardless of the presence or
absence of cold AMV-cDNA. The observation that the extent
of PolI hybridization at completion in the presence of AMV
competitor was about 40"ó of that in the absence of competi-
tor (Figure 3.3) supports the view that PolI- and AMV-cDNAs
truly represent different length copies of the same nRNA
population.
Reassociation of Keratin Alt{V-cDNA and PolI-cDNA5
to Chick Ery throcyte DNA
Figure 3.4 shows the reassociation kinetics of
keratin AMV-cDNA and PolI-cDNA in the presence of a vast
excess of chick erythrocyte DNA. Keratin AI'4V-cDNA reasso-
ciated in a biphasic curve with major transitions at Cot1,
values at about 9 x 101 and ? x 103 nol.s.1-1 when reasso-
FIGURE 3.4.
REASSOCIATION OF KERATIN AMV- AND POLI -CDNAS WITH
CHICK ERYTHROCYTE CHROMOSOMAL DNA
Chick DNA (average length 0. 5 kb) 'hlas nixed with kera-
tin AMV- or PolI-cDNA, denatured at 100oC and allowed to
renature at 60oC. Aliquots were removed at appropriate
intervals and the extent of renaturation estimated using St
nuclease (high salt buffer) and HAP. Normalization factors
are shown.
(a) St nuclease assays in high salt buffer;A keratin AMV-cDNA (0%), tr keratin
PolI-cDNA (10%), . chick erythrocyte DNA,
labe1led by nick translation (L4%).
(b) HAP assays; a keratin AMV-cDNA
E keratin PolI-cDNA (II%),
erythrocyte DNA (AZOO) (18%).
(10%),
o chick
0
Ito
.!IJotttttúo
É,
-eo
20
40
60
80
20
I 00
40
60
80
100-t 430 1
Log
2
Cot (mol.s.lll)
a
ooA
tr
D
A
b
oAA
oA
a
58.
ciation was measured by resistance of the cDNAs to digestion
with S, nuclease (Figure 3.a @)). In contrast, less than
L0% of keratin PolI-cDNA had reassociated by a Cotli of
9 x 101 mo1.s.1-1 rnd there was only one major transitionwith a ro"r4 of 9.5 x LOZ mol.r.1-1, a value indistinguish-
able from that of the unique sequence fraction of chick ery-
throcyte DNA (Figure 3.a G)) .
When assayed on HAP, as shown in Figure 3.4(b),
keratin Alt{V-cDNA reassociated in a broad curve, indicating
the presence of both reiterated and unique sequences. In
some experiments, the curve appeared to be biphasic, but
insufficient points made it difficult to make accurate
estimates of the Cotk for the transitions. A biphasic
curve was previously reported by Kemp (1975) for the AMV-
cDNA reassociation reaction. The Cot6;for the transitions
were estimated. as 7 x 100 and 9 x L02 mol.s.1-1. Keratin
PolI-cDNA reassociated with a single transition at a ,otU
of about 4.7 x ir}Z mol.,.1-1, a value again indistinguish-
able from that of the unique sequence f.raction of chick ery-
throcyte DNA (Figure 5.4(b)). The differences in reasso-
ciation of AMV- and PolI-cDNAs were reproducible with dif-
ferent batches of each cDNA and a]-so when the cDNAs were
each annealed with the same batch of sheared chick erythro-
cyte DNA, ruling out the possibility that the differences
r{rere caused by variations in the fragrnent size of the chick
erythrocyte DNA.
6. Reassoci-ation of Elongated PolI-cDNA to Chick
Erythrocyte DNA
If PolI- and AMV-cDNAs were different length
59.
copies of the same nRNA species, then elongated PolI-cDNA
should show sinilar reassociation kinetics to AMV-cDNA.
PolI-cDNA elongation was performed as described by crawford
and Wel1s (1976) and the nethod is outlined in Figure S.5.3H-1"b"11ed Polr-cDNA was hybridized to keratin rnRNA and the
hybridized cDNA was used as primer for AMV reverse trans-
criptase in a synthesis mix containing all 4 deoxyribonucleo-
side triphosphates unlabelled. After reverse transcription,the elongated probe with a labelled 3r end and an unlabelled
5r section was separated fron the nRNA by alkali digestion.Radioactively labe1led products ranging in length from 0.2
to 0.65 kb were obtained, as estinated frorn their velocityof sedimentation through linear rjeo to 40% sucrose gradients
relative to PolI- and AIvtV-cDNA markers (Figure 5.6). Three
size classes of increasing length r4rere selected from these
gradients and the HAP assayed Teassociation kinetics of
each of these were compared with those of the starting PolI-
cDNA (Figure 3.7). From Figure 3.7 it is clear that the
apparent rate of reassociation increased with increasing
length of the probe. In fact, the curve obtained with el-ongated PolI-cDNA of average length 0.55 kb was almost in-distinguishable frorn that for the AMV-cDNA reassociation
reaction (Figure 3.4 (b) ) .
The radioactive labe1 is confined to the 5 ' 150
nucleotides of these elongated molecules and since the PolI-
cDNA before extension only showed unique reassociation kin-
etics (Figure 3.4(b)) the sequence of cold nucleotides added
by the AMV polymerase, as directed by the nRNA template,
must represent a reiterated sequence in the chick genome.
FIGURE 5.5
SCHEMATIC REPRESENTATION OF THE ELONGATION OF
POLI-CDNA WITH AMV POLYMERASE
For experimental detailsSee text for descriPtion.
see Crawford and We1ls (1976).
Poly A
5'+3'mRNA+ oligo dT$'-ì/^zrrr 5r
Hybrid ize
5t VUv:3, Hybrid3t-<.'1u\^ 5r
Poll cDNA(3rf laoettea
Reverse Transcriptasecold d NTPs
Alkali digestion
3'l-- 5'unlabelled labelled
FIGURE 3.6.
SELECTION OF SIZE CLASSES OF ELONGATED POLI-CDNA
Elongated keratin PolI-CDNA was centrifuged through
L0-40e0 sucïose gradients in 10 mM Tris-HCl, PH 7.4, 1 nM
EDTA, 0.1 Ir{ NaCl at 57r000 ï.p.m. for 16 hours at soc using
a Beckman SW41 rotor in a Beckman L5-50 preparative ultra-
centrifuge. Gradients were fractionated into 0.5 nl
fractions using an ISCO density gradient fractionator, and
10 pl aliquots from each fraction l\rere counted in Toluene
Triton scintillation fluid to detect elongated nolecules.
AMV- and PolI-cDNAs were centrifuged on paral1e1 gradients
to provide nolecular l^ieight markers, and their banding
positions weïe determined as described above. The bars
indicate fractions which were pooled to give the different
size classes; I PolI-cDM; o AI4V-cDM; O elongated Po1I.
Sedinentation is from right to 1eft.
1 220 Bases average length.
Z. 350 Bases average length.
3. 550 Bases average length.
=o.o
.Lîã
200
100
0.55 kb 0.15 kb
2 1
15 10
Fraction NumberSedimentation
25 20 5 1
FIGURE 3.7 .
REAS SOCIATION OF ELONGATED KERATIN POLI-CDNA
TO CHICK CHROMOSOMAL DNA
All assays were carried out on HAP. The average size
of the elongated PolI-cDNA size classes was estimated from
their velocity of sedinentation through linear L0-40%
sucrose gradients relative to AMV-cDNA and PolI-cDNA markers
run on paral1el gradients (Figure 3.6). Normalization
factors are shown.
o PolI-cDNA (10%) ' E 220 base elongated
PolI-cDNA (0e"), I 350 base elongated PolI-cDNA
(8%), . 550 base elongated PolI-cDNA (2f%).
0
L¡o
.g(,o.tthtúoÉ,
s
20
40
60
80
1001 4320 1
Log Cot (mo¡.s.¡.-1)
60.
These results suggest that keratin nRNA has a
3t end of 150 nucleotides representing unique sequences in
the chick genome, covalently linked to a reiterated sequence
located further towards the 5r end of the messenger.
7 Thermal Stabilities of Reassociated Duplexes of
Keratin AN[V- and PolI-cDNA
In this section, the therrnal denaturation studies
described for AMV-cDNA reassociated duplexes were carried
out by Kernp (Figure 5.8(a) and (b)).
As shown in Figure 3.8(a), the thernal stabili-
ties of the keratin AMV-cDNA duplexes changed narkedly with
Cot, two major components with Trs of about 73"C and 90oC
being evident. Replots of these curves (Figure 3.8(b))
demonstrate clearly that more than 60% of the duplexes with
T-s less than 85"C, which were present at C^t 1.9 x 101 mol.-m- - -._1
s.1 -, had transferred into duplexes of high thernal stability
by Cot L.L4 x 104 nol. s.1-1. In contrast, there was a rnuch
srnaller change in thermal stability of keratin PolI-cDNA
duplexes with Cot (Figure 5.8(c)) and no significant transfer
of duplexes from low thermal stability to high thermal
stability (Figure 3.8(d)). One najor component with u T,
of approxirnately 83"C was evident, although in addition
there were some duplexes with Trs lower than 70oC-
8 Buoyant Densities of Reassociated Duplexes of
Keratin AMV- and PolI-cDNAs in the Presence of
Actinomycrn_q
Studies on the buoyant density in actinonycin D
CsCl of native DNA fragments containing sequences comple-
FIGURE 5.8.
THERMAL STABILI-TIES OF KERATIN AI{V- AND POLI-CDNA
REASSOCIATED DUPLEXES
Reaction mixes identical to those described in Chapter
III.B.3.(a) r4rere incubated to the various Cot values shown
and the thernal stabilities determined (Chapter III.B.3.(b).
(a) Keratin AMV-cDNA duplexes after incubation to
Co, values (nol. s.1-1) of ^ 1.9 x 10-1,
r 5.7 x 10
L.L4 x 104 .
3tr 3.75 x 10^
2,
(b)
(c)
The same data as in (a) re-plotted after multi-
plying each value by the normalized % of total
cDNA which bound to HAP at 60oC.
A
Keratin PolI-cDNA duplexes after incubation to
Cot values (nol. s.1-1) of ^ 4.g x LOz ,
1.6 x 10
g.3 x ro2, o 3.r x LoS,4
(d) The same data as in (c) re-plotted as in (b).
Data in Figure 5.8(a) and (b) hlas obtained by Kenp.
I
NB
ba
À
dc
80
60
40
200
80
60
40
20
60
20
o
o
i
P
iocl
80
60
îtr¡¡
;xit,
claot'Bo
40
20
40
70 80 90 ?0 80 90 100100
Temp.oC
61
mentary to keratin AMV-cDNA (Lockett and Kernp, 1975) dernon-
strated that this DNA contained G + C rich sequences. The
density of duplexes of keratin AMV-cDNA and PolI-cDNA, formed
by reassociation of the cDNAs with chick erythrocyte DNA to
high Cot, \^Ias therefore investigated. The duplexes r\rere
treated with St nuclease before actinomycin D-CsCl gradient
centrifugation in order to remove single-stranded regions,
nisnatched regions and oligo-dT tails on the cDNA rnolecules.
The main band of reassociated chick erythrocyte
DNA had a density of 1.590 gm/cc on actinomycin D-CsC1 grad-
ients (Figure 3.9). Keratin AMV-cDNA duplexes banded at an
average density of about 1.566 gm/cc (Figure 3.9(a)). In
contrast, keratin PolI-cDNA duplexes banded at an average
density of 1.582 gn/cc (Figure 3.9 (b) ) . Since actinonycin
D binds preferentiaLLy to G + C rich sequences (Kersten,
Kersten and Szybalski, 1966) resulting in a displacernent of
such sequences to the light side of CsCl gradients, keratinAMV-cDNA rnust contain a G + C rich sequence which is not
present in keratin PolI-cDNA. Although there was slight
variation in the apparent density of the main band DNA in
independent experiments, the variation in density difference
between main band DNA and cDNA duplexes was 0.002 gm/cc or
less in 5 independent experiments. When the chick erythro-
cyte DNA was replaced with calf thyrnus DNA, no radioactivity
above background was detectable on the gradients (results
not shown). Since the duplexes had been treated with
nuclease St prior to centrifugation, hle assume that the
apparent densities of the cDNA duplexes would not have been
significantly affected by the presence of single-stranded
or mismatched regions.
FIGURE 5.9.
CENTRIFUGATION OF KERATIN AMV- AND POLI-CDNA
REASSOCIATED DUP LEXES ON ACTINOÌ\4YCIN D-CSCl GRADIENTS
(a)
Duplexes formed at Cot 2.0 x 104 rnol.s.1-1 *ut"
treated with S, nuclease and then centrifuged on actinonycin
D-CsCl gradients as described in Chapter III'B'4'
Keratin AMV-cDNA duplexes (c.p.m. ) ,
chick erythrocyte DNA (AZOO).
(b)
o
a
o
o
Keratin PolI-cDNA duplexes (c.p.n.),
chick erythrocYte DNA (AZOO).
1.5901.582b rl
1.590
tt'566
I5
4
a500
400
300
100
100
50
3
2
I
=ÈrJ
200otoN
5
4
=.o-
o2
o(o¡
3
I
015
Fraction10
Number
525 20
62.
D DI S CUSSI ON
1. Analysis of rnRNA Preparations and cDNAs
Although keratin nRNA preparations have been
previously reported to be greater than 95eo pure by the cri-
teria of ce11-free translation and forrnanide gel electro-
phoresis (Partington et a1., L973; Kernp et a1 ., L974b; Kenp
et aI., L974c) the data in Figure 5.1 indicate that as much
as 30% of arly given preparation night be ribosomal breakdown
products. Most of the ribosonal contaminant appeared to be
breakdown material fron 18S rRNA. Tables 3.1 and 3.2, how-
ever, provide evidence that the cDNA transcribed from these
messenger preparations contained insignificant amounts of
ribosomal reverse transcríPts, regardless of whether AMV
reVeISe transcriptase or PolI was used for the synthesis.
The facts that both of these enzymes required oligo-dT to
prime cDNA synthesis (Table 5.1) and AMV- and PolI-cDNAs
averaged 0.55 and 0.15 kb in length respectively (Figure
3.2), suggest that these two cDNAs represent different
length copies of the keratin nRNA, both copies initiating
at the 3t poly(A) tract. The competition hybridization
experiments shown in Figure 3.3 clearly demonstrated that
polI- and AMV-cDNAs share common sequences and hence that the
two enzymes did not copy different sets of the nRNA popula-
tion. All of these resultsrtaken together, suggest that
AMV- and PolI-cDNAs truly represent different length copies
of the keratin nRNA population. The AMV-cDNA replesents a
0.55 kb copy of the messenger starting at the 3t poly(A)
tract, while PolI-cDNA is representative of the 150 nucleo-
tides at the 3t end of the keratin nRNA'
63.
? Reassociation and Thermal ProP erties of AMV-cDNA
and PolI-cDNA Duplexes
The data in Figure 3.4 indicate that the kinetics
of reassociation of AMV-cDNA and PolI-cDNA to a vast excess
of chick erythrocyte DNA were quite different. The AMV-
cDNA showed a broad transition, indicative of the presence
of reiterated and unique sequences, while the PolI-cDNA
reassociated at a rate indistinguishable from that of unique
sequences.
The reassociation kinetics of AMV-cDNA have been
previously explained (Kernp, 1975) by proposing that each
keratin AMV-cDNA molecule contained a region of very sinilar
sequence which could form poorly rnatched duplexes with the
corïesponding regions of other non-identical keratin genes'
The reiterated kinetics 'hleïe attributed to cross-hybridiza-
tion since they were not apparent when the reassociation
reaction was assayed under Very stringent low salt St assay
conditions (Kernp, 1975). As the assay conditions became
less stringent (high salt st assay, HAP assay, Figure 3.4)
the reiterated kinetics became more obvious. Another
interpretation, however, which is indistinguishable fron
this mismatching hypothesis on the presently available data,
would propose that each AMV-cDNA molecule contained a short
faithfully conserved region which could form short perfectly
natched duplexes with the corresponding regions of non-
identical keratin genes. This sequence will be denoted
the repetitive sequence in the sense that it must occur many
times in the chick genome, while avoiding detailed descrip-
tion of the nature of this repetitive sequence (i'e', whether
64.
it is a short common repeat or a longer mismatching repeat.
Figure 5.8(a) and (b) indicate that the duplexes formed early
in the AMV-cDNA reassociation were of low thermal stability,
a result in complete agreement with the hypothesis that the
reiterated sequence is either short or mismatched. The
observation that the same cDNA rnolecules present in these
low melting point duplexes early in the reactions were later
transferred into duplexes of much greater thernal stability
can be explained by proposing that each keratin AMV-cDNA
molecule also contains a sequence which has diverged to the
extent that it cannot cross-hybridize with non-identical
keratin genes and, therefore, behaves as if it is unique in
the genome. 0n the assumption that the nultiple keratin
genes have arisen by gene duplication, this nodel would
predict that different regions of the keratin genes have
diverged to varying extents, the reiterated regions showing
1ittle divergence while the unique regions have diverged to
a much greater extent.
The behaviour of keratin PolI-cDNA observed here
was entirely consistent with this nodel since they behaved
as if they contained only unique sequences. It follows
that unique sequences must be located adj acent to the poly (A)
tract of keratin nRNA. The reassociation kinetic data
obtained using elongated PolI-cDNA as probe would then
suggest that the unique sequences at the 3t end of the mRNA
rnolecule are covalently linked to a reiterated sequence
further to the 5r end of the molecule.
An alternative model nay also be considered in
relation to the data. In this nodel it is assumed that
65.
the degree of sequence divergence between different keratin
nRNA species is distributed randornly along the molecules.
This "random divergence" nodel can account for the observed
behaviour of keratin AMV-cDNA:DNA duplexes since cDNA can
be displaced from a duplex by other molecules, provided
that more stable cornplexes can be formed (Becknann and
Daniel, L974). In fact, the latter process can be invoked
to explain the high thermal stability of the hybrids formed
late in the reaction using either of the two models.
The two models lead to different predictions
about the behaviour of PolI-cDNA however. Whereas the unique
and repetitive sequence model predicts all the observed pro-
perties of the PolI-cDNA molecules, the random divergence
model predicts, in principle, that the PolI-cDNA should be
equally as repetitive in nature as the AMV-cDNA. While the
latter was clearly not the case, the interpretation is com-
plicated by the fact that short duplexes are intrinsically
less stable than longer duplexes. It is, therefore, poss-
ible that mismatched duplexes of PolI-cDNA did not forrn
because their thermal stability was too low relative to the
incubation tenperature. A quantitative evaluation of this
possibility is necessary in order to distinguish between
the two models.
Southern (1975a) demonstrated that restriction
endonuclease fragrnents of mouse satellite DNA ranging in
length frorn 120 to 600 bases could all reassociate to form
misrnatched duplexes. The depression of the Trnt of these
duplexes from those of the native rnolecules due to mis-
matching was constant, irrespective of the length of the
66.
fragments. The depression of Trs of the shortest reassocia-
ted duplexes relative to long native duplexes could be
accounted for aS the sum of the depression due to chain
length and the constant nisnatching factor of 6oC. These
data can be used to predict the thermal stability of the
keratin PolI-cDNA duplexes expected for the random diver-
gence model. The maximum T, of keratin PolI-cDNA duplexes
was lower by 7oC than that of the corresponding AMv-duplexes
(Figure 3.8).
The therrnal stability of globin PolI- and AIW-
cDNA hybrids has been compared (Kenp et al. , in preparation)
and it was found that the PolI hybrids had a Tn 4oC lower
than the AMV hybrid. This reduction in T,n was presumed to
be due to the shorter length of the hybrid. of the 7oc re-
duction in T, for the keratin PolI-cDNA duplexes when com-
pared to the AMV-cDNA duplexes, 4oC can presunably be accoun-
ted for by the reduced length of the duplex while the re-
3"C can be attributed to the difference in G + C
of the PolI- and AIW-cDNA duplexes (Figure 5.9)
of nismatched PolI-cDNA duplexes should therefore be
7"C below those of mismatched AMV-cDNA duplexes (72-73"C;
Figures 4 and 7) or about 65-66oC. Clearly, since the
reaction was done at 60oC, the rate of fornation of many of
these duplexes would be negligible (Bonner et a1. , L973) .
However, because of the broad range of T,ns exhibited by
mismatched AMV-cDNA duplexes (Figure 3.8(a) and (b)) about
half of the mismatched PolI-cDNA duplexes would be expected
to have T s distributed between 65oC and 75oC. Suchm
duplexes should form at a tate comparable to those of the
mismatched AMV-cDNA duplexes (Bonner et a1., I973) after
maining
content
TheTsm
67.
correcting for the effect of length on rate.
The random divergence model, therefore, Predicts
that the assay of keratin PolI-cDNA reassociation kinetics
on HAP should result in a complex reassociation curve show-
ing both unique and reiterated characteristics. Since PolI-
cDNA was observed to reassociate with unique kinetics, it
was concluded that the randorn divergence model is incompat-
ible with the data and hence that it is most probable that
keratin rnRNA contaíns distinct regions of unique and repeti-
tive sequence. This conclusion depends on the validity of
the assumption that the depression of the T, of a duplex due
to random rnisrnatching is independent of length, âû effect
that has apparently only been shown for mouse satellite DNA.
3 Model for the Sequence Organi zation of Keratin
nRNA
The data available suggest a nodel for the
sequence organization in keratin nRNA (Figure 3.10). The
experimental data, assumptions and deductions upon which
this model is based, are surnmarized be1ow.
(a) The length of keratin nRNA is about 0.8 kb
(Kemp et al. , L974b), it contains a 3t
poly(A) tract of average length 60 nucleo-
tides and a 51 7 methyl G cap structure
(Morris and Rogers, in Press) .
At least 150 bases adjacent to the poly(A)
tract are unique sequences (present results).
We assume that the actual length is 150
bases and designate this region as 3t
(b)
(c)
(d)
68.
unlque sequences.
About 50-50% of the length of each keratin
AMV-cDNA molecule, 550 bases long, is com-
posed of repetitive sequences, since in
separate experiments 30-50% of the radio-
activity in these molecules appears as a
fast transition resistant to nuclease St
at high salt concentration when reannealed
with chick erythrocyte DNA (Figure 3.a@)).
The reassociation kinetic analysis of
elongated PolI-cDNAs of varying lengths
(Figure 3.7) also suggests that the repe-
titive sequence is fairly long and extends
to within close proximity (100 nucleotides)
of the 3t unique region of the nRNA. The
presently available data do not a1low any
further conclusions to be made about the
nature , size or position of this repeti-
tive sequence. Most or aL1- of the repe-
titive sequences are covalently attached
to unique sequences since elongated PolI-
cDNA shows reiterated reassociation kinetics
when reannealed with chick erythrocyte DNA
(Figure 3.7) .
Low melting point duplexes of the repeti-
tive region were depressed in T, by 18-20oC
from well-rnatched duplexes (Figure 5.8 (a)
and (b)). Therefore, the sequences are
mismatched at about 20% of the bases
(e)
(f)
69.
(Bonner et ãT., L973) and hence have
diverged by about l}eo in sequence (Southern,
1971). Alternatively the repetitive re-
gion is very short and highly conserved.
Keratin chains are about 100 amino acids
long (0'Donnel1, 1973b; Walker and Rogers,
I976a). They do not appear to be synthe-
sized as larger polypeptide precursors,
since the molecular l^ieight of keratin
translated fron keratin nRNA in both rab-
bit reticulocyte and wheat ernbryo ce11-
free systems was identical to that of
native keratin and the N-terminal sequence
was identical (Partington et ãI., T973;
Kenp et al. , L974c). Therefore the kera-
tin coding sequence is about 300 bases
1ong.
(g) Cornplete amino acid sequences have been
deternined for feather keratin chains from
other species (0'Donne11, 1973b; OtDonnell
and Inglis, L974) and partial sequences of
tryptic peptides from enbryonic chick
feather keratins have also been determined
(Walker and Rogers, I976b; Kenp et al.,
1975). While regions of about 10 amino
acids at the N- and C-termini vary consi-
derably in sequence, the internal section
of about 80 arnino acids is highly conserved
in the keratin chains which have been
(h)
70.
sequenced. The limited sequence studies
on ernbryonic chick feather keratins in this
internal region demonstrated that the
sequences of 4 different N-terminal tryptic
peptides were identical in the region from
amino acids L2-28 except for a single pro-
line + serine substitution in one of the
four sequences. In the absence of further
sequence data we conclude that the amino
acid sequence of the internal region of
all embryonic chick feather keratins are
very similar. If these regions ltlere
identical in amino acid sequence then they
could differ by a rnaximum of about L7% in
their nRNA sequence (Kohne, 1970). From
(e), if the low melting point of early
duplexes hlas due to nisrnatching then the
repetitive sequences in the keratin nRNA
would have diverge d by about l0% , a value
cornpatible with a suggestion that the re-
petitive sequence in keratin rnRNA is in
the coding sequence.
From the amino acid composition of feather
keratin (Walker and Rogers, 1976a) it can
be predicted that the DNA coding for ker-
atins should have a G + C content of about
54.7%, compared with 4L% for total chick
DNA. Such a G + C rich region is
present at a distance greater than 150
bases from the poly(A) region, again
7L.
suggesting that the coding sequence is in
this region.
Thus it appears that keratin mRNA contains unique
and repetitive sequences covalently linked. The 3t 0.15 kb
of the nRNA constitutes at least a portion of the unique
sequence, while the reiterated sequence is located further
towards the 5 I end of the nRNA molecule. This reiterated
sequence could be relatively long and mismatching oT short
(about 50 bases) and faithfully conserved. The apparent
increase in the rate of reassociation of elongated PolI-cDNA
with increasing length, however, favours the misrnatching
interpretation. On the basis of the sequence hornology of
different feather keratin proteins, it is ternpting to specu-
late that the reiterated sequence corresponds to the keratin
coding sequence
FIGURE 5.10.
MODEL FOR THE ARRANGEMENT OF SEQUENCES IN KERATIN NRNA
The 150 nucleotides adjacent to the 3t poly(A) tract
constitutes a unique untranslated sequence. Further to
the 5 I end of the mRNA and covalently linked to this 3t
unique sequence is a repetitive sequence (short and faith-
fully conserved or longer and mismatching) , a G + C rich
sequence, the coding sequence and presurnably a 5r untrans-
lated sequence. The currently available data pernit no
further conclusions about the sequence arrangement within
this 5r region of 550 bases.
The orientations and approximate sizes of AMV- and
PolI-cDNAs are also shown.
150base s
60ba se
mRNA 5'
Polr cDNA
AMV cDNA 3
550 bases
coding,repetitive, G +C rich and 5'
untranslated sequences?
.3'unrque &
u ntrans -lated
3'
Poly A
5'550 bases
CHAPTER IV
SEPARATION OF RIBOSOMAL AND KERATIN SEQUENCES
BY PHYSICO-CHEMICAL METHODS
7?
CHAPTER IV
SEPARATION OF RIBOSOMAL AND KERATIN SEQUENCES
BY PHYSICO-CHEMICAL METHODS
A INTRODUCTION
This chapter describes attempts to obtain a partial
purification or quantitative separation of chick keratin and
ribosornal sequences. The proposed hybrid-density gradient
procedure for the investigation of gene arrangement relies
on the density difference between RNA:DNA hybrids and single-
stranded DNA in CsTSOO (see Chapter V.A). To generate
hybrid without aIly DNA reannealing, hybrid formation is
carried out in vast RNA excess. For the nethod to be able
to contribute infornation about gene arrangement, there are
two basic requirements:
That the sequences being studied constitute a
high enough proportion of the total DNA being
used in the experiment to allow their detection.
That the RNA being used to form hybrids is pure
or that the DNA used in hybrid forrnation is
devoid of those sequences which are complemen-
tary to the rnaj or contaminating RNA species '
Fron the Co"ra for total chick DNA (Kernp, 1975), ít can
be calculated from Laird (1971) that the genome size of chick
is about L.2 x L¡LZ daltons. This corresponds to 1.9 x 106
kb of DNA per genome. Reassociation kinetic analysis has
revealed that there a¡e about 100 ribosomal genes in the
chick genome (P. Krieg, personal communication). since the
1
2
73.
amount of DNA coding for mature rRNA species per ribosomal
cistron is about 7 kb, the total length of sequence coding
for mature rRNA per genome is about 700 kb. From this it
can be calculated that the sequences coding for mature rRNA
species constitute about 0.037% of the chick genome, a
figure consistent with that calculated by Sinclair and Brown
(1971). This repïesents a sufficiently large proportion of
the genome to permit ribosomal sequences to be readily de-
tected without any pre-selection for these sequences. Cot
analysis has also shown that there are about 100 genes coding
for keratin in the chick genome (Kenp, 1975). The length
of keratin nRNA is about 800 bases (Kernp et aI . , L974b) , 60
of which make up the poly(A) tail at the 3t end of the mole-
cule (Morris and Rogers, in press) . Thus the total length
of DNA complemerLtary to keratin sequences in the chick gen-
ome is about 74 kb. It can therefore be calculated that
keratin sequences constitute 0.0059% of the chick genome;
about l/r'th the amount attributable to ribosornal sequences.
From Chapter III.C.1 it is known that ribosomal se-
quences are the major contaminating RNA species in keratin
nRNA preparations. While it is possible to remove some of
these contaminating species by oligo-dT cel1u1ose chromato-
graphy of the nRNA, some ribosomal sequences remain adsorbed
to the column and elute with the messenger. Although the
keratin nRNA has a lairLy long poly(A) tract, for some
unknouln reason only I\eo to 30% of the total nRNA loaded on
to oligo-dT cellulose binds to it. The unbound naterial
has been shown to contain about 70% of the loaded keratin
messenger as determined by in vitro translation (B'C' Powell,
personal communication). The hybrid-density gradient pro-
74.
cedure, however, requires large amounts of highly purified
nRNA and any attempts to remove contaminants from keratin
nRNA preparations results in large losses of nessenger.
Added to this, the 1ow yields of keratin nRNA from arly given
preparation (20-50 ug) and the 1ow proportion of the genome
coding for keratin, made it necessary to investigate the
possibility of separating genomic ribosomal and keratin
sequences with a view to enriching for keratin sequences.
It has been known for rnany years that the sequences
coding for 18S and 28S rRNA in chick have a G + C content of
64% (Sinclair and Brown, 1971). While keratin genes are
also G + C rich (54.5% G + C) compared with total chick DNA
(|Le,) , they appear to be associated with DNA of lower G + C
content (Lockett and Kernp, 1975) . The possibility of ex-
ploiting the different G + C contents of keratin and ribo-
somal sequences to obtain a quantitative separation of these
sequences was, therefore, investigated. It has been re-
ported (G. Partington, personal communication) that Xenopus
globin sequences can be purified 10-fold by thermal elution
from HAP using DNA of length 10 kb. since DNA of higher
molecular r,rleight was required for the proposed hybrid-
density gradient rnethod for the analysis of gene organization,
thermal elution studies were perforrned using DNA of single-
stranded size >, 50 kb. The fractionation of ribosomal and
keratin sequences on CsCl gradients was also studied using
native DNA of high single-stranded rnolecular weight. It
is known that when high molecular weight native DNA is
fractionated on CsCl density gradients, the G + C rich kera-
tin sequences band with the bulk of chick DNA as a result of
their association with sequences of lower G + C content
75.
(Lockett and Kemp, 1975). This effect may perrnit a greater
separation of keratin and ribosomal sequences than would be
e4pected on the basis of the G + C content of their coding sequences
a1one. After quantitative separations of ribosomal and
keratin sequences by the cscl density procedure, keratinsequences could be further purified by cycles of cscl grad-
ient centrifugation in the presence of actinonycin D and
netropsin sulphate, as previously described (Lockett and
Kemp, 1975).
B METHODS
1 Therrnal Elution fron HAP
(a) In aqueous buffer
Using a glass column surrounded by a water
jacket linked to a Haarke waterbath, HAP (prepared as
described in Chapter II.B.5) was washed and equili-brated with 0.LZ M phosphate, pH 7.0 at 60oC. High
nolecular l^Ieight DNA (prepared as described in Chap-
ter II.B.4) (>.50 kb) in 2 mI of the same buffer was
loaded on to the column, allowed to equilibrate to
60oC for 2 min and then passed through the HAP under
slight pressure. The column was washed a further 2
times with 2 mL of buffer equilibrated to 60oC. All60oC eluates r^rere pooled and stored. The equilibra-tion temperature was raised in 5oC steps to 100oC and
5 washes, with 2 mL of 0.I2 M phosphate (0.18 lul Na+),
pH 7.0, were carried out at each ternperature. Afterthe 100oC wash, the column was eluted with 0.4 M
(0.6 M Na+) phosph ate, pH 7 .0, at 70oC to remove arLy
76.
rernaining DNA. Eluates
perature were pooled and
frorn each equilibration tem-
stored separately.
Fractions were cleared of HAP fines by
centrifugation in an MSE Superminor for 5 min at max-
imum speed. The DNA content of each eluate was deter-
mined fron AZOO measurement using a Zeiss spectrophoto-
meter.
(b) In buffer containing 50% fornamide
The procedure was exactly as described
above except that DNA was loaded on the column at 40oC
in 0.LZ M phosphate, pH 7.0, containing 50% formanide
(Ajax Chenicals Ltd., Laboratory Reagent grade).
Elutions were carried out in 5oC steps up to 95oC
using this same buffer as described above. Any re-
maining DNA was rernoved with 0.4 M phosphate, pH 7.0,
containing 50% fornamide at 60oC. The DNA content of
each eluate was determined by measurement of AZg0
(since formanide absorbs at 260 nm but not at 280 nn).
? CsCl Eq uilibrium Density Gradient Fractionation
8.53 g Of CsCl was dissolved in a minimum volume
of TE (Chapter II.B.4) and the solution made up to a total
weight of 15.5 g with high molecular weight DNA solution
(100, 200 or 400 ug in TE). This gave a final gradient
volume of 9 n1 and a density of I.700 g/cc. Gradients I'\Iere
covered with paraffin oi1 and centrifuged in a Becknan nodel
L2-50 preparative ultracentrifuge using a fixed angle Ti50
rotor at 20"C for 60 to 70 hours at 32,000 r.p.m.
77.
Gradients were fractionated from the botton into
23 drop (0.5 nl) fractions using a Gilson Microfractionator
rnodel FC-80E. The density gradient achieved was deterrnined
by reading the refractive index of every third fraction.
Where 100 ug of DNA was fractionated on a gradient the A260
of the fractions 'hras deternined using a Zeiss spectrophoto-
meter. Where 200 ug and 400 ug of DNA was fractionated,
fractions were diluted 2 and 4 times respectively with water
before A determination.260
3 Detection of Ribosomal and Keratin Sequences
All fractions from a given experiment were made
up to the same DNA content with E. coli DNA. Fractions
were made 0.1 M with respect to NaOH to denature the DNA.
They were neutrali.zed by the addition of 0.1 volumes each of
1 M Tris-HCl pH 7.5 and 1 M HCl, made up to 10 x SSC by the
addition of an equal volume of 20 x SSC and the DNA immo-
bili zed by filtering through nitrocellulose discs using a
Universal membrane filter apparatus. Filters were washed
twice with 20 ml of 6 x SSC, cut in half, dried and baked
at 80oC in vacuo for at least 2 hours. Half filters r,rlere
incubated in 2 x SSC containing 1 x Denhardt solution (0.02%
each of Fico11, BSA and polyvinylpyrrolidone' see Denhardt,
1966) at 65oC for at least 6 hours, then blotted dty. Dry
filters r^Iere stacked in a siliconized scintillation vial
containing saturating amounts of ]-25T labelled 185 and 285
ribosonal RNA (Chapter II.B.7) or 32p labelled keratin cDNA
in 2 x SSC, 0.5% SDS, pH 7.2, and hybridization carried out
for 20 hours at 65oC. Half filters l^Iere given 3 x 8 hour
washes in 2 x SSC, 0.5% SDS, at 65oC. Filters were then
78
rinsed in 70%
bound to each
ethanol, dried at 65oC and the radioactivity
filter determined.
Cs C1
tion
When 200 or 400 Ug of DNA was fractionated on
gradients, U or 4 respectively of each gradient frac-was treated in the manner described above.
C RESULTS
1 Thermal Elution fron HAP
Figure 4.L shows the results obtained from aîa-
lytical thermal elution studies. The bulk of the high
molecular hreight DNA appeared to denature with " T, of 9SoC
in aqueous buffer (Figure a.1 (a) ) . In this same aqueous
buffer, keratin sequences denatured with a T, of 97oC while
at 100oC only 25% of the ribosomal sequences could be eluted
from the column. It was possible, however, to elute the
remaining ribosomal sequences with 0.4 M phosphate, pH 7.0,
at 70"C. These results suggested that, by carrying out
preparative thermal elution at 95oC in aqueous buffer, 70%
of the keratin sequences would remain bound to the HAP along
with only 232 of the total chick DNA. If the bound mater-
ial , after elution at 95oC with 0.LZ M phosphate, pH 7.0,
was eluted with 0.4 M phosphate, pH 7.0, at 70oC, this mat-
erial would be enriched 5-fold for keratin sequences, but
4-fo1d for ribosomal sequences as well. Thus, to obtain
enrichment of keratin sequences without enrichment for ribo-
somal sequences by this procedure, thernal elution should
be carried out at 95oC to remove most of the chick DNA and
again at 100oC to elute 70eo of the keratin sequences, but
only 25% of the ribosomal sequences. This would decrease
FIGURE 4.L..
FRACTIONATION OF KERATIN AND RIBOSOMAL SEQUENCES
BY THERMAL ELUTION FROM HAP
100 ug 0f DNA was bound to HAP and eluted with phosphate
buffer (0.18 M Na+) at increasing tenperatures using aqueous
buffers and buffers containing 50% fornanide as described in
Chapter IV.B.1.(a) and (b). The DNA samples from each
elution r^rere denatured, neutralized and bound to nitrocellu-
lose discs as described in Chapter IV.B.5. Filters hlere
cut in half and challenged with either ¡3U1-keratin AIr4V-cDM orT25r labelled rRNA. The points plotted for aîy given elu-
tion temperature represent the counts hybridized at that
ternperature and al-l lower elution temperatures, expressed as
a percentage of the total counts hybridized after all elu-
tions had been performed.
(a) In aqueous buffers; o total chick DNA
(AZOO) , o keratin sequences, ^ribosomal sequences.
In buffers containing 50% formamide; o
total chick DNA (A2gg), o keratin sequences,
a ribosomal sequences.
(b)
'Eo
:c¡trGIL+.ØgE)
U'
o
80
60
40
100 a
60 70
100 b
80
60
40
r"Bå p. oc 90 100
20
tto!(úL+,Ø
oE)
.=Ø
oã\.
20
60 70Tem P.
40 50 oc 80 90
79.
the enrichment for keratin sequences to 2.9-fold, but this
fraction would show no significant enrichment for ribosomal
sequences over unfractionated DNA. By elution with 0.4 M
phosphate, pH 7.0, at 70"C after the 100oC thernal elution,
a vast purification of ribosomal sequences could be obtained.
From Figure a.2@), however, it is apparent that, ãt the high
temperatures necessary for the partial purification process
in aqueous buffers, gross nicking of the DNA occurs.
Since this breakdown hras undesirable for the proposed hybrid-
density gradient system, analytical studies on thernal elu.
tion were repeated using buffers containing formamide.
Figure 4.1(b) shows the thernal elution data
using 0.LZ M phosphate, pH 7.0 in 50% formanide. Under
these conditions total chick DNA exhibited a T, of 77"C,
keratin sequences showed a T, of 78.5"C while ribosomal
sequences denatured with a T, of 83.5oC. These data sug-
gested that, under these conditions, it was not possible to
obtain any significant partial purification of keratin
sequences from total chick DNA. It should be possible,
however, to obtain a significant purification of ribosomal
sequences. If the double-stranded fraction was eluted with
0.4 M phosphate, pH 7 .0 in 50qz fornanide at 60oC after ther-
na1 elution at 80oC with 0.LZ M phosphate, pH 7.0 in 50%
formamide, it would contain 75% of the ribosornal sequences
and only L5% and 25% of the total chick DNA and keratin
sequences respectively. This would correspond to a 5-fo1d
enrichment of ribosomal sequences over total chick DNA.
The purification of keratin sequences from
chick DNA using thernal elution in fornarnide contain-to ta1
FIGURE 4.2.
(a)
(b)
EFFECT OF TEMPE RATURE AND HAP CHROMATOGRAPHY ON
SINGI,E -]STRANDED DNA MOTECULAR ITEIGHT
DNA (1 ug) of single-stïanded síze >- 50 kb in 0.5 nl of T'E'
buffer was heated at the ternperature indicated in the
figure for 5 rnin. Samples r^rere chilled on ice, eth-
anol precipitated, redissolved in 2.5 pl of 0.1 M NaOH,
10 rnM EDTA and electrophoresed in a 0.4% alkaline aga-
rose ge1 as described in chapter v.8.1.(b) (ii). À Phage
DNA (46 kb), 186 phage DNA (50 kb) and À DNA digested
with EcoR, (chapter II.B.9) were loaded in 0.1 M NaOH,
10 mM EDTA and used as rnolecular weight markers. The
size of the starting chick DNA is also shown. The gel
was stained with ethidiun bronide and photographed as
described in Chapter V.8.1.(b) (ii). O, origin'
chick DNA (1 ug), before and after therrnal elution at
80oC in buffer containing 50% formanide, was electro-
phoresed in a 4% alkaline agarose 9e1, the ge1 was
stained with ethidiun brornide and photographed. Due
to the lower voltage at which this gel was electrophor-
esed, the strands of À, 186 and the 20.8 kb EcoR, frag-
ment of À DNA have separated. O, origin'
N.B.: Owing to the inadequacies of photographic reproduction
and the broad distribution of DNA, it is difficult to see
the DNA in the track showing chick DNA after thernal elution.
The DNA was in fact distributed from about 20 kb to less
than 1 kb.
kb4630
20.8
7.2
5-6
30
20.8
\R. \DNA
80.
ing buffers was nowhere neaT as great as that obtained using
aqueous buffers at higher temperatures. However ' the fact
that 75% of the ribosomal sequences could be removed from
the bulk of the DNA, which harbours the keratin sequences,
and the observation that lower temperatures lead to less
thermal degradation of the DNA (Figure a.2@)) suggested
that thermal elution on a preparative scale under these con-
ditions should be investigated.
Table4.lshowstheyieldsfromapreparative
thermal elution. Figure 4.2(b) shows the electrophoretic
rnobiLity of the DNA in alkaline agarose gels before and
after thermal elution relative to markers consisting of
phage À DNA, phage 186 DNA and À DNA digested with EcoRt.
It is apparent from this figure that chick DNA of initial
length >, 50 kb is broken down to a range of fragments vary-
ing from less than 1 kb to Z0 kb. Cornparison of the electro-
phoretic pattern of chick DNA fragments after treatment at
800c (Figure 4.2(a)) with the pattern after HAP chromato-
graphy at 80oc in the presence of fornamide (Figure 4 -2(b))
indicatesthat additional breakdown, over and above that
caused by temperature, occurs during the chromatography step'
It should be pointed out that the gels in Figure a.z@) and
(b)weTeelectrophoresedatdifferentvoltages.Atthe
lower voltage used in FiguÏe 4.2(b) the strands of À,186
and the 20.8 kb EcoR, fragment of I DNA have separated while
no such separation was observed at the higher voltage (Figure
a.2@)).
Thehybrid-densitygradientprocedurerequrres
single-stranded DNA size classes ranging fron 5 kb to 50 kb
TABLE 4.L
RECOVE RY OF DNA FROT'4 PREPARATIVE THERMAL ELUTION
IN THE PRESENCE OF FORMAMIDE
Treatment Amount
900 ug
7þzþE
200 ug
Recover
0 .83%
t) ) 0,LL.L 'O
DNA loaded
0.I2 M phosphate, pH 7.0,50ro formamide, 80oC
0.4 tr{ phosphate, pH 7 .0,50% formamide, 60oC
Total recovery 23%
500 ug Of high molecular weight DNA (> 50 kb) was
loaded on to each of 3 HAP columns, packed volume 2 nls.
5 ml Of 0.I2 M phosphate (0.18 M Na+), pH 7.0 in 50% forma-
mide was equilibrated to 80oc and passed through the HAP.
This I^Ias repeated twice with 2.5 nl volumes of the same
buffer. 5 m1s Of 0.4 M phosphate (0.6 M N"*), PH 7'0 in
50% formamide was equilibrated to 60oC and passed through
the HAP. This was repeated twice with 2.5 ml volumes of
the same buffer. Total low and high salt eluates were
dialysed against 2 changes of T.E. and ethanol precipitated'
Precipitates l^Iere redissolved in T.E. and DNA content
estinated from AZOO readings.
81.
as discussed in the next chapter. From Figure 4.2(b) it
is apparent that the higher molecular weight classes could
not be obtained from DNA which has undergone thermal elution.
Since the recovery of DNA from this procedure was low and
further losses were expected to occur on selection of the
single-stranded size classes (Chapter V), the technique was
of no practical use for the quantitative separation of ker-
atin and ribosomal sequences.
CsCl Gradient Centrifusation')
It has been known for many years that the ribo-
somal genes of chick are G + C rich and band to the heavy
side of the main band of chick DNA in a CsCl equilibriun
density gradient (Sinclair and Brown, 1971). While keratin
genes are also G + C rich, they have been shown to band
directly under the main peak of chick DNA in CsCl gradients
when the DNA being studied is of high molecular ltleight
(Lockett and Kemp, 1975). It therefore seerned that CsCl
gradient centrifugation rnight allow the partial purification
of ribosomal sequences and the separation of ribosomal and
keratin sequences on a preparative basis.
Increasing amounts of Dl\Ir\ (> 50 kb) were centrifuged in
CsCl and the banding position of ribosomal and keratin
sequences determined in each case by filter hybridization
(Figure 4.3). When 100 ug of DNA was centrifuged to equili-
briun in CsCl (Figure 4.3(a)), 80% of the ribosomal counts
appeared as a single sharp band on the heavy side of the
gradient around fraction 9. The rest of the ribosomal
counts appeared as a lower density shoulder trailing toward
the chick DNA rnain band. The rRNA, iodinated for use as
FIGURE 4.3.
CsCl GRADIENT FRACT TONATION OF KERATIN AND
RIBOSO}4AL SEQUENCES
Different amounts of native chick DNA of single-
stranded size >. 50 kb were centrifuged to equilibriun in Cs-
Cl gradients as described in Chapter IV.B.2. Gradients were
fractionated, the DNA frorn each fraction innobiLized on
nitrocellulose discs, the discs split in half and one set of
half discs challenged with L25I rRNA and the other with 32p
cDNA to keratin nRNA.
(a)
(b)
(c)
100 pg DNA; o
A L2'5r c.p.m
o 32p c.p.m.
200 ug DNA; o
A r25r c.p.n32^o P C-p.m.
AZOO (total chick DNA),
(ribosomal sequences),
(keratin sequences).
AZOO (total chick DNA),
(ribosonal sequences),
(keratin sequences).
Azoo (total chick DNA),
(ribosomal sequences),
(keratin sequences).
400 ug DNA;
r25 _
^'lc
o
p m
32po c.p.m.
Density increases frorn right to left with the main band of
chick DNA having a density of I.700 g/cc. Densities were
deterrnined bY refractometrY.
N.B.: 1. 32p cDNA used in this experiment was prepared
using a 60 nin incubation at 37"C (see chapter II.B.6.(a)).
2. . CsCl gradient centrifugation is not expected to
lead to afly degradation of the DNA (Tomizawa and Anraku,
1e6s).
o å
32P
cpm
x 1
o-2
A z
so ? è
ET
oNS
CD
OÀ
¡ -232
P "p
'n10
o N
32P
cpm
x 1
o-2
1251
cpm
x l0
-3
ooP
ôÀqr
<.)
A z
soA
zoo
oq, o o 3
o@ o o 3
Þ
oo (D
o ñe
@ o o J
o cÊ (ì
o ñ
Þ¿
ô6 o z 3sl
ET o
o or a
grC
'I(t
l
ê!¡ o õ' )
п
ô€l
õ'
zà 3 ET o
z¿ Cgt o
N o N (tt
{ o t
N o
-{ ¡
¡O
aJr
o xl\)
oO
l\,S
CD
æo €
l\) (nE
'O
o
f\)
(.)
]\)Þ
O¿
N)C
.tSgl
Ctl
o1\
)
1251
cpm
x
10-2
t25¡
sprn
x l0
-2
82.
the probe, was obtained from snall and large ribosomal sub-
units and the RNA species were subsequently purified as
described in Chapter II.B.5. Nevertheless, the counts in
the trailing shoulder could stil1 be due to contaminating
non-ribosomal sequences in the radioactive probe. Alterna-
tively, the trailing edge could be due to some non-specific
aggïegation artefact brought about by the high viscosity of
the high molecular weight DNA in the main band of chick DNA
(Brown and Weber, 1968a). Keratin sequences appear to band
at two positions. one peak, carrying 80% of the counts,
appears directly under the nain band of chick DNA as expected,
while the smaller peak bands along with the ribosomal se-
quences. While it is possible that keratin genes exist in
two different density classes ' a more 1ike1y explanation for
this bimodal distribution is that 20% of the cDNA, in this
particular prepaTation, u/as copied from contaminating rRNA
in the keratin nRNA preparation since the cDNA synthesis mix
was incubated at 37"C for t hour (See Chapter II.B.6. (b)).
These data indicated that keratin and ribosomal sequences
could be separated by CsCl density gradient centrifugation'
The next problem was to determine how much DNA could be
loaded on a single gradient without adversely affecting the
resolution.
Whenzo0ugofDNAwasusedinthedensitygraa-
ient analysis (Figure 4.3(b)), a secondary peak of ribosomal
hybridi zatiort appeared underneath the main band of chick
DNA. Under these conditions, the secondary ribosomal peak
(fractions L2 to 15) constituted about 12% of the total
ribosomal counts, while the distribution of keratin counts
between main band and ribosomal peaks remained the same as
83.
for Figure a .3 @) .
Figure a.3 (c) shows the profiles obtained when
400 ug of DNA was centrifuged in a CsCl gradient. The
secondary peak of ribosomal hybridization, first observed
when 200 ug of DNA was subjected to CsCl density analysis,
has become more prominent accounting for 37% of all the
ribosomal counts, suggesting that it is probably a viscos-
ity artefact. Again the distribution of keratin counts
between main band and ribosomal peaks renained the same as
before.
From Figure 4.3(b) and 4.3(c) it appears that
the rnaximum amount of DNA r,vhich can be fractionated on a
single CsCl gradient, without inviting artefacts caused by
viscosity effects, is 200 ug. Under these conditions
(Figure 4.3(b)) fractions 5 to 10 contained 72% of the
ribosomal counts and only 9.4eo of the total chick DNA. By
pooling these fractions, therefore, it should be possible
to obtain a 7.7-Ïo1d enrichment of ribosomal sequences over
total chick DNA. The rest of the DNA would contain 80% of
the keratin sequences and only L8% of the ribosomal sequences.
D DISCUSSION
The results shown in Figures 4.L and 4.3 indicate that
keratin sequences cannot be effectively separated from the
bulk of chick DNA by either thernal elution fron HAP (Figure
4.1) or CsCl density gradient fractionation (Figure 4.3).
Ribosomal sequences, however, can be purified 5-fo1d and
7 .7 -fold using these rnethods. More irnportant than the
partial purification of ribosomal sequences, however, is
84.
the fact that
separation of
these procedures can provide a quantitative
keratin and ribosomal sequences.
1 Thermal Elution
From Figure a.1 (a) it can be calculated that
therrnal elution carried out in aqueous buffers, ernploying
elution steps at 95"C and 100oC, would a1low a 2.8-fo1d puri-
fication of keratin sequences. In addition, the DNA of
this keratin-containing fraction would only contain 25% of
the total chick ribosomal sequences. Figure 4.2(a), how-
ever, shows that gross breakdown of the DNA occurs during
these thernal elution steps when carried out at such high
temperatures.
When DNA nelting experiments are performed in
buffers containing formamide, the T, of the DNA is, theore-
tically decreased by 0.72"C for each L% of fornanide in the
buffer (McConaughy et aI., 1969). To reduce the amount of
thernal degradation occurring during the thermal elution
study, buffers containing 50eo formamide weTe used. Figure
4.1(b) shows that while no significant separation of keratin
sequences from total chick DNA could be achieved under these
conditions, it was possible to obtain a 5-fold purification
of ribosomal sequences. After therrnal elution at 80oC, the
single-stranded fraction contained 75% of the keratin se-
quences and only 25eo of the ribosomal sequences. Thus no
enrichnent for keratin genes was possible using thernal
elution in the pïesence of formamide, but a quantitative
separation of keratin and ribosomal sequences could be ob-
tained. Less thermal degradation of the DNA occurred under
these formamide conditions than in aqueous buffers at higher
85.
temperatures (Figure 4.2) .
From Figure 4.I it is apparent that there is a
greater difference in T, between keratin sequences and total
chick DNA in aqueous buffers than in buffers containing for-
mamide. This is presumably an artefact caused by the dif-
ference in single-stranded DNA molecular weight at the dif-
f erent ternperatures (Figure 4 .2 (a) ) s ince , if all other
variables were constant, the presence of formamide would be
expected merely to displace both the total chick and keratin
Sequence nelting profiles to a lower ternperature range with-
out changing the difference between their Trt. Poorly
matched DNA duplexes aTe known to adsorb to HAP, but dif-
ferent explanations for the mechanisrn by which single-
stranded DNA nolecular weight could affect the relative Trt
of keratin sequences and total chick DNA nust be invoked
depending on the nature of the DNA:HAP interaction.
(a) If DNA with any vestige of double-stranded
structure adheres to HAP
Actinonycin D-CsCI gradient analysis has
revealed that keratin genes have a higher G + C con-
tent than their surrounding DNA (Lockett and Kemp,
1975). Thus, regardless of the length of DNA con-
taining keratin sequences (within reason), the T, of
these sequences should be dictated by the T, of the
structural genes. If the total chick DNA has regions
of G + C rich sequences, then the probability of arly
particular fragrnent containing such Sequences increases
with increasing DNA sîze. Thus, with increasing
single-stranded DNA size, the apparent Tr of total
86.
chick DNA would be expected to increase. If this is
the mode of action of HAP, then the observed thermal
elution results would suggest that G + C rich
sequences are present on all chick DNA fragments of
lengths 15 kb to 40 kb. (This is the size range of
fragments after incubation at 80oC (Figure a.2@) -)
(b) If the proportion of the molecule in
double-stranded form is the inportant
cr iter ion for DNA bindins to HAP
Since DNA is badly degraded at the high
temperatures required for therrnal elution in aqueous
buffers, keratin genes would constitute a large pro-
portion of arLy fragment containing keratin sequences.
Thus, the T, of such fragments would be dictated by
the T_ of the keratin gene, and the resolution betweenm
this value and the T, for total chick DNA would reflect
the difference in their G + C content. At the lower
temperatures required for thernal elution in the pre-
sence of formamide, however, the DNA is subjected to
less thermal d.egradation and hence keratin genes might
constitute too sma1l a proportion of the duplex to
perrnit it to adhere to HAP when the rest of the mole-
cule is in single-stranded form. Thus, where the
DNA is of relatively high single-stranded size, the
melting characteristics of keratin sequences would be
dictated by those of the surrounding sequences which
are of lower G + C content (Lockett and Kernp' 1975).
If this is the node of action of HAP and the above
explanation of the T* anomaly is correct, then the Tt
87.
data would inply that each keratin gene
with a large length of Dl,lA of lower G +
This would suggest that, if the keratin
linked, they are either associated with
vening sequences of lower G + C content
spacers of lower G + C content.
is associated
C content.
genes are
large inter-
or large
Comparison of the Tn for total chick DNA Ín
the pïesence and absence of formamide (Figure 4.1) indicates
that the apparent T,n depression per T% formamide in the
elution buffer, is considerably less than 0.72"C (McConaughy
et à1., 1969). There aTe two possible explanations for
this .
(1) The arrangement of G + C rich sequences in
chick genome may cause high rnolecular
weight DNA to exhibit an abnormally high
T, when assayed on HAP, as described in
( a ) above. This interpretation assumes
that these G + C rich sequences constitute
a sufficiently high proportion of the DNA
fragrnent to allow the fragment to remain
adsorbed to the HAP. If this combined
effect of G + C rich sequence distribution
and single-stranded DNA molecular weight
is the reason for the unexpectedly small
effect of fornanide on T* in these studies,
the nelting curve in the presence of for-
rnamide would be expected to be steeper
than that observed using aqueous buffers.
This effect is not obvious in Figure 4.L.
88.
(Z) There may be some interaction between
single-stranded DNA and HAP in 0.LZ M
phosphate, pH 7.0; it must be independent
of the fornanide content of the buffer and
the magnitude of the interaction would
have to increase with increasing single-
stranded DNA size. As discussed earlier,
the DNA obtained from thermal elution
studies in the presence of formamide is of
higher single-stranded molecular weight
than that obtained in aqueous buffers.
In this situation the interaction of the
DNA with HAP would be greater when T*
analysis is carried out using buffers con-
taining forrnarnide. Higher temperatures
would therefore be required to elute these
structures from the column and this would
lead to an anomalous increase in the appar-
ent T of the DNA.m
Neither of these explanations can, by itself,
account satisfactorily for the abnormal nelting character-
istics of total chick DNA observed during denaturation arLa-
lysis on HAP using buffers containing formamide. It is
quite likely that all the factors nentioned above play some
part.
When preparative thernal elution was attenpted
using buffers containing formamide, the yields of DNA were
very low (Table 4.1). This observation, along with the
fact that the DNA obtained fron the thernal ftactionation
89.
proceduïe did not contain molecules of the largest size
class necessary for the hybrid-density gradient procedure
(Figure 4.2(b)) , indicated that the nethod was inadequate
for the separation or partial purification of keratin and
ribosomal sequences.
') CsCl Gradient Fractionation
cscl density gradient centrifugation allowed a
7.7-foLd purification of ribosornal sequences from total
chick DNA but no enrichment for keratin sequences (Figure
4.3). The bulk of the DNA left after removal of the ribo-
somal peak contained 80% of the keratin sequences and only
18% of the ribosomal sequences. For preparative separation
of ribosomal and keratin sequences using DNA of length >. 50
kb, however, a maxirnum of 200 Ug of DNA could be loaded on
to each gradient. The relative abundance of ribosomal
Sequences in the chick genome rendered it unnecessary to
parti aLLy purify these sequences prior to their use in the
characterisation of the proposed hybrid-density gradient
procedure. On the other hand, keratin sequences are pre-
sent at a 10-fold lower level than ribosornal sequences and
So enrichment for these Sequences might be neces saTy before
keratin gene arrangement could be analysed by the hybrid-
density gradient technique. The CsCl fractionation des-
cribed was to be used as the first step of this partial
purification, providing a quantitative separation of keratin
and ribosomal sequences. Keratin sequences could than be
further purified by cycles of CsCl density gradient centri-
fugation in the presence of antibiotics actinomycin D and
netropsin sulphate (Lockett and Kernp, 1975)'
90.
E. APPENDIX
The conclusions made in this chapter regarding the
proportions of keratin and ribosomal sequences present in
different fractions are only valid if the radioactive probe
used in the filter hybridization r.tras present in saturating
amounts. In order to investigate this, 100 ug of chick DNA
hras imnobilized on filters as described in Chapter IV.B.3
and the filters challenged with increasing concentrations ofr25r labelled 28s rRNA and 3H labelled. keratin cDNA respec-
tively. Typical results for these saturation hybridization
analyses are shown in Figure 4 .4 .
From these results it appeared that the levels of the
different radioactive probes necessary for the saturation of
100 ug of chick DNA on a filter were 500 ng/rnl for 28S rRNA
and 60 ng/mL for keratin cDNA. In accordance with these
findings then, these were the concentrations of the radio-
active probes used in the experiments described in Chapter
IV. Saturation filter hybridizations were not performed
for r25r labelled 18s rRNA, but in all experiments, equal
concentrations of L25I labelled 185 and 285 rRNA were used.
Since 28S rRNA and 18S rRNA were isolated from large and
sma1l ribosornal subunits respectively and re-purified by
sucrose gradient centrifugation, the two rRNA species should
show little cross contarnination. Thus the bulk of the I25I
labelled 18S rRNA should be 18S sequences rather than 28S
rRNA breakdown products. In this situation, given that 18S
rRNA is slightly less than half the length of 28s rRNA, the
hybr idization conditions used for L25I 1abelled 18S rRNA
would be expected to be saturating.
FIGURE 4,.4.
CONDI TIONS FOR SATURATION OF FILTER HYBRID.IZATIONS
chick DNA (100 ug) ü.as irnnobilized on nitrocellulose
discs. Filters were challenged with increasing amounts ofL25r Labelled z8s rRNA or 5H l-abelled keratin cDNA. Fil-
ters were washed extensively in 2 x SSC, 0.5% SDS, dried
and counted.
(a) Filter hybridizations using increasing arnounts
- L25-or r labelled 28S rRNA.
Filter hybridizations using increasing arnounts
of 3H keratin cDNA.
(b)
20
10
o a
400 500 600
a
a
o
o
oo
NIo
x
EÀo
rt)N
00 100 200
ng
300
R NA /mr26
24
22
20
18
f6
o
ob
NIo
x
=Ào
(r'
14
12
10
I6
4
2
t
60
ng
80
cDNA /ml0 20 40 100 120 140 160
CHAPTER V
THE HYBRID-DENSITY GRADIENT PROCEDURE FOR THE
ANALYSIS OF GENE ARRANGEMENT
91.
CHAPTER V
THE HYBRI D-DENSITY GRADIENT PROCEDURE FOR THE
ANALYSIS OF GENE ARRANGEMENT
A INTRODUCTION
This chapter deals with studies on the isolation of
relatively homogeneous size classes of single-stranded DNA
and their use in the prelininary investigation of a density
gradient nethod for analysing the arrangernent of eukaryotic
gene s .
The technique is a slight modification of the pre-
hybridi zation technique used to study the arrangement of the
sequences coding for 18S and 28S rRNA of Xenopus laevis
(wallace and Birnstiel, L966; Birnstiel et al., 1968; Brown
and Weber, 1968b) and the IRNA genes of this toad (Clarkson
and Birnstiel , Ig73). It relies on the density difference
between single-stranded DNA and RNA:DNA hybrids. Figure
5.1 outlines the principle of the technique. Consider the
tandemly repeated gene farnily shown. The structural gene
gives rise in vivo to a nRNA species which can then be iso-
lated and hybridized to single-stranded DNA of specific size
classes. This pre-hybrtdization step must be carried out
with a vast excess of RNA so as to drive RNA:DNA hybridiza-
tion ahead of DNA reassociation. when the resultant struc-
tures aTe centrifuged in CsTSOO gradients, the position at
which the hybrid bands, relative to the two extremes of 100%
hybrid and single-stranded DNA, should be dependent on the
proportion of the DNA nolecule in hybrid form. Hybrids
made using DNA of single-stranded size much larger than that
FI GURE 5 .1.
SCHE},,{AT I C REPRESENTATION OF THE PRINCIPLE BEHIND
THE I{YBRID-D ENSITY GRADIENT PROCEDURE FOR THE
ANALYSIS OF GENE ARRANGEMENT
structurallgene lspacerl
.vvvv-,w mRNA
gene length
----Repeatedgene family
Singlest randedDNA sizeclasses
Single StrandedDNA
Top
1
2
3
4
> gene lengthrepeat
unit lengthmultiple repeatunit length
Hybridize with vast excess of RNA
æ1
2
3
4
21oo
(,goE
$-oLo.oE
z
av\/\/\tyy\,
¿\t\r.,/vt aw\tl Nvvvrúv M
100%fvnrid
1
CIFO4 s rad ient
3+4
Bot tom Density
o?
of a single repeat unit should have a density identical to
that of hybrids generated using DNA of repeat unit length
if the genes are tandenly linked. If the genes being
studied aïe not tandernly linked, the density of the hybrid
generated should approach that of single-stranded DNA as
the size of the DNA used in hybrid fornation increases.
For tandenly linked repeated gene fanilies, it should be
possible, by this technique, to determine the average length
of a ïepeat unit by knowing the length of the RNA gene pro-
duct and using single-stranded DNA, which is expected to be
much larger than the repeat unit in the pre-hybridization
reaction. This would, of course, require a knowledge of
the relationship between the density at which a hybrid bands
and the proportion of the DNA in hybrid form relative to the
two extremes mentioned above. The repeat unit length ca1-
culated in this way could be confirmed by carrying out simi-
1ar hybrid-density analyses using single-stranded DNA of
size classes increasing up to and beyond the repeat unit
length calculated from the initial investigation. Analyses
of these ð.ata in a manner similar to that used by Lizardi
and Brown (1975) to deternine the length of the silk fibroin
gene using CsCl gradient centrifugation in the presence of
actinornycin D, should al1ow the accurate deternination of
the ïepeat unit length for tandenly linked repeated gene
families.
The technique ulas to be characterized using the ribo-
somal cistrons of chick as a nodel system. When perfected,
it was to be used for the analysis of keratin gene arrange-
ment since it is believed that the keratin genes of chick
constitute a tandernly repeated gene farnily (Lockett and
93.
Kernp, 1975). The requirements for the characterization of
the sys tern were :
1. The isolation of homogeneous single-stranded DNA
size classes.
2. The ability to detect the generated hybrids in
the densitY gradient.
3. To obtain complete RNA:DNA hybridization without
DNA reassociation during pre-hybridization'
4. To empirically determine the relationship
between the density of the hybrid formed and the
proportion of the DNA rnolecule in hybrid form,
relative to the two extremes of single-stranded
DNA and 100% hYbrid.
It should be stressed that this project was conceived
and had reached aî advanced state of developrnent before
restriction enzyme analysis and recombinant DNA technology
becarne readily available to Australian molecular biologists.
Indeed, Australiats first recombinant DNA facility was only
built in late 1977. These studies were continued in the
belief that the results obtained by this approach would pro-
vide independent information about average gene arrangement
which could then be critically compared with that obtained
from restriction or recombinant DNA analysis. During the
course of this study, the restriction enzyme map of the
chick ribosomal genes was published (Mcclernents and skalka,
Lg77) perrnitting a critical evaluation of the hybrid-density
technique.
94.
B METHODS
1 Isolation of SP ecific Size Classes of Single-
stranded Chick DNA
(a) Alkaline sucrose gradient centrifugation
EthanolpÏecipitated,highnolecularweight
chick DNA was dissolved in 0.1 M NaOH, 10 mM EDTA at
a concentration of about 1 ng per nl. 5 -20% Sucrose
gradients made up in 0.1 M NaOH, 10 mM EDTA, 0'9 M
NaCl, were overlayed with 500 pl of the DNA solution
and centrifuged in an sw41 rotor using a Becknan L5-
50 preparative ultracentrifuge for 4 hours at 57r000
r.p.rn. at 15oC. Gradients were fractionated by up-
ward displacement using an ISCO density gradient
fractionator, model 640.
Fractionswereneutralizedbytheaddition
of 0.1 volumes each of 1 M Tris-HCl, PH 7'5, and 1 M
HC1, diluted 2-foLd with sterile bidistilled water and
precipitated with 2.5 volumes of ethanol at -20"c for
a mininum of 2 hours.
(b) Alkaline ag.arosesel electrophoresis
Alkaline agarose gel electrophoresis was
essentially as described by McDonell et al. (1977) .
(i) Preparative
20 cm x Z0 cm x0.5 cilr 0'5% Agarose
gels in 30 nM NaOH, 2 nM EDTA' were poured
either verti calLy between glass plates or hori-
95.
zorLtaLLy inside 0.5 cn high borders. The cir-
cuit between the electrophoresis tanks contain-
ing 30 nM Na0H, 2 mM EDTA and the gel was com-
pleted by wicks made from Whatmann 5 MM paper
soaked in electrophoresis buffer . 0.2-Z ng 0f
randomly sheared chick DNA in 2 mI of 0.1 M
NaOH, 10 nM EDTA, 5% glycerol, containing bromo-
phenol blue, was loaded on top of the gel (or
in the loading slot of a horizontaL ge1) and
electrophoresis performed at 100 volts overnight
or at 300 volts for 6 hours.
(ii) Analytical
20 cm x 20 cm x 0.2 cm Horizontal
0.4% agarose gels in the buffer described above
were used for this purpose. 1 ug Sanples were
loaded in the loading slots in 40 pl of 0.1 M
NaOH, 10 mM EDTA, 5% glycerol, containing brorno-
phenol blue, and electrophoresis carried out at
250 volts for 4 to 6 hours.
After electroPhoresis, the gel was
neutralized by soaking in 0.5 M Tris-HCl, PH
6.8 for 10 min, stained with a 0.01% solution
of ethidium bromide in bidistilled water for 15
nin and destained in mono-distilled water for
30 nin. The destained gel, illuninated with
ultra-violet light in an 0liphant ultra-violet
viewing box, I^ras photographed with a Nikornat
E.L. camera fitted with a red filter using
96.
Kodak recording film 2475. Filns were developed
with Kodak HC-110 develoPer.
Whole À phage DNA (gift 0f
Dr. R.C. Crawford), whole phage 186 DNA (eift of
R.B. Saint) and I DNA digested with EcoR, restric-
tion endonuclease were used as molecular I^Ieight
markers in these analytical gels allowing mole-
cular weight estimation of fragments varying in
size between 3.3 and 45 kb. New digestions of
À DNA with EcoR, were carried out for each ana-
lytical gel so as to mininize any nicking of the
DNA on storage.
(c) Extraction of sing 1e-stranded DNA fron
preparat ive alkaline agarose se 1s
After electrophoresis of the DNA in a
preparative alkaline agaïose gel, the gel was sliced
into 1 cm thick strips at right angles to the direc-
tion of electrophoresis. Three nethods for the
extraction of the DNA from these strips were investi-
gated.
(i) Solubilization of agarose with KI
followed by HAP ch romatog raphy
Agarose striPs üIere broken into
smal1 pieces by passage through a syringe and
were dissolved by the addition of I.25 volurnes
of a saturated solution of KI (SSKI) followed
by incubation at 37"C for 30 nin. solubilized
agarose strips were then loaded on to an HAP
q7
column of 2 mI packed volume, previously equili-
brated at 37"C with SSKI, equilibrated to 37"C
and passed through the HAP under slight pressure.
The columns were washed twice with 5 ml of SSKI
and twice with 5 rnl of 10 nM phosphate, pH 7.0,
to remove arly remaining agarose. The DNA from
the solub i1-ized strips was then eluted with
0.4 M phosphate, pH 7.0. The 0.4 M phosphate
eluate was centrifuged to remove any HAP as
described in Chapter IV.B.1.(a) , dialysed
against 2 changes of T.E., each of volune 4
litres. The amount of DNA remaining after this
procedure was deternined by estimation of the
AZOO. The DNA was concentrated by ethanol pre-
cipitation (ChaPter II.B.1) .
(ii) Krsradient centrifugation
Agarose slices were solubiLized as
described above and neutralized by the addition
of 0.1 gel volumes of 1 M Tris-HCl, PH 7'5, and
0.03 gel volumes of 1 M HCl. Ethidiurn bromide
was added to a final concentration of 50 ug/nl'
The solution was adjusted to a refractive index
of L.420 (density about 1.5 g/cc), and centri-
fuged at 40,000 r.p.m. for 40 to 60 hours at
15"C using a Ti50 rotor and Beckman LZ-50 pre-
parative ultracentrifuge (Wolf, 1975) ' After
centrifugation, the position of the DNA band
was deterrnined frorn the fluorescence of the
bound ethidiun bronide when viewed under ultra-
98.
violet light of wavelength 2573 Å. The DNA
was removed by side puncture of the centrifuge
tube using an 18 gauge needle attached to a 2
nl syringe. Ethidiun bronide was removed from
the DNA by 3 extractions with isoamyl alcohol
while KI was renoved by 2 dialyses against 4
litres of T.E. The amount of DNA remaining
after this treatment was determined by reading
the AZOO and DNA was concentrated by precipita-
tion with 2.5 volumes of ethanol at -20"C over-
night.
[ 111J Centrifusation of the agarose slice
Whole agarose slices hlere placed in
screwcap 50 rn1 centrifuge tubes and centrifuged
at 191000 r.p.m. for 20 min at 4oC in a Beckman
JAz0 rotor using a Beckman J-218 preparative
centrifuge. The supernatants were decanted,
neutral ized by the addition of 0.1 volurnes of
1 M Tris-HCl , PH 7.5, and 0.03 volumes of 1 M
HCl and concentrated by precipitation with Z '5
volumes of ethanol at -20"C for a ninirnum of 2
hours.
2 Pre - hyb ridi zation Conditions
20 to 100 ug of DNA frorn a given single-stranded
size class in 1nl of T.E. hlas nade up to 10 nM EDTA, 0.1 M
NaOH. The DNA solution was neutralized by the addition of
0.05 volumes of 1 M Tris-HCl, PH 7.5, and 0.1 volumes of 1 M
HCl and placed on ice. RNA, which had been denatured by
99.
boiling, r{ras added to the DNA and the solutions mad.e up to a
final volume of 2 ml by the addition of o.z ml of z0 x ssc
and 0.65 ml of water. rn the earry experiments, z0 ug of125r laberled zgs rRNA or z0 ug each of L25r rabelled rBS and
28s rRNA of specific activity about z x 105 c.p.m./pg was
added to the pre-hybri dization rnix which was then incubated.
at 60"c for 60 nin. under these conditions, the RNA drivenhybridization reached a Roa of a.Lz5 nol.s.1-1 with DNA re-association reaching a cot of 0.r25 nol.s.1-1 to 0.625 rno1.s.
1-1. These values should be compared with the st assayed.
ooa2 for the hybridi zation of z8s cRNA to its cDNA of 1 x L}-z_1no1.s.1 - (Figure S.1(a)) and the St assayed. ,orb for the
ïeassociation of chick ribosomal sequences of 10 mol.s.1-1(P. Krieg, personal communication) . pre-hybrid.i zation r^ras
also carried out using unlabelled rRNA under these same
conditions. In later experiments, 60 ug of 2gs rRNA alone
or ó0 ug each of 18s and 2BS rRNA was hybridized to zo ug
of DNA at 60oc for 50 min in a total volume of z mL. und.er
these conditions a Rot of about 0.2 mol.s.1-1 rrrd a coa ofabout 0 . 06 mo1 . s . 1- 1 r", achieved. Hybr ið,izati-on was
stopped b¡r chilling the reaction mix on ice.
3. Generation of 100% Hybrid
A 50 U1 solution containing keratin nRNA (Z ug)
and 32p labelled cDNA made to keratin mRNA (5 x 105 c.p.rn.)in hybridization buffer (10 rnt'4 Tris-HC1, pH 7.0, 1 mM EDTA,
0.18 M Nac1, 0.05% sDS), overlayed with paraffin oil, r^ras
boiled for 2 min then incubated at 60"c for t hour. The
hybridi zation solution was then added to medium salt stassay buffer (0.03 M Na-acetate, 0.15 M NaCl, 0.001 M ZnSO4,
5% glycerol, pH 4.6) containing Tz ug of denatured coalfish
100.
DNA, 4 units of 51 nuclease was added and S, digestion was
allowed to proceed at 37"C for 30 min. The St digestionI^ras stopped by the addition of 0.02 volumes of 25eo SDS.
5 ug of E. coli tRNA was added as carrier and the s, nuclease
resistant hybrid precipitated with 2.5 volumes of ethanol at
-20"C for at least 2 hours. Hybrids were redissolved inT.E. About 100,000 c.p.n. of hybrid was added to Z mI ofa solution identical to that used for pre-hybridizationprior to CsrS0O gradient centrifugation. L00% Hybrids
using rRNA and cDNA made to rRNA (Chapter II.8.6.(b)) rrras
prepared in the same r4ray.
4 . Cs ^SO,-¿-+
Gradient Centrifugation
High purity grade CsrS0O was obtained from
Kawecki Berylco Industries Inc. and recrystalLized from boil-ing water. 5.63 g 0f recrystaLLized CsTSOO was dissolved
in the ninimum volume of water. The pre-hybridization
solution was mixed with the CsrS0O solution and made up to
a final volume of 9 ml (density 1.5 e/cc). The solutionwas transferred to siliconized centrifuge tubes and the
gradients centrifuged at 28,000 r.p.m. for 60 to 70 hours
at 20"C in a Ti50 rotor using a Beckman L2-50 preparative
ultracentrifuge. Gradients were fractionated fron the
bottorn into 0.3 n1 fractions using a Gilson micro-fraction-
ator and densities determined by refractometry. The amount
of DNA in each fraction was determined by measurement of the
AZOO of the gradient fractions.
5. Detection of Hybrids
Three nethods for detection of hybrid were
101 .
employed.
(a) RNArase resistance of iodinated hybrid
Hybrids hrere forned using an excess ofr25r labelled rRNA to drive the reaction. Afterfractionation of the CsTSOO gradients, 1.7 rn1 of 10
nM Tris-HCl , pH 7.5, 4 rnM EDTA containing pancreatic
ribonuclease A (Sigrna) at a concentration of 7 1tg/nL
r^ras added to each fraction. RNAiase digestion was
allowed to proceed at 37"C for 30 min and the amount
of radioactivity resistant to RNArase in each fractionwas deternined by TCA precipitation.
(b) Bindine of iodinated hybrid to nitro-cel1ulose
This was perforrned essentially as described
by Wallace and Birnstiel (1966). The method relieson the observation that RNA:DNA hybrids bind to nitro-cellulose filters while RNA does not.
Hybrids were forned using an excess ofL25r labelled rRNA. After fractionation of the grad-
ient, 150 U1 of each fraction was added to 6 rnl of2 x SSC and heated at 60"C for t hour. Fractions
rtrere chilled on ice, filtered through nitrocellulosediscs and both sides of each disc were washed with
50 ml of 2 x SSC. Filters rtrere dried at 65oC and
the radioactivity bound to each filter determined.
r02.
(c) Immobilization of DNA on nitrocellulosefollowed by filter hybridization
The third method was essentialLy as des-
cribed by Birnstiel e_t 41. (1968) . Unlabelled RNA
was used in the pre-hybridization step and, afterCsrS0O density gradient fractionation, the fractions'hrere diluted to 1 rnl by the addition of water and. nade
0.1 M with respect to NaOH. Fractions r^rere left in0.1 M NaOH at room temperature for at least t hour toallow denaturation of DNA:RNA hybrids and extensive
alkaline hydrolysis of the RNA. Neutralization,innobilization of DNA on nitrocellulose filters, prê-
treatment of filters and conditions of hybridizationüiere as described in Chapter IV.B.3 with the exception
that one set of half filters was challenged with I25I
label1ed rRNA to detect the hybrid, the other set
being challenged with 3H labelled cDNA made to rRNA
to allow detection of the banding position of single-stranded ribosomal DNA sequences.
6. Determination of Single-stranded Sequence Densit
In method (a) and (b) for hybrid detection, the
AZOO peak was used as the marker for single-stranded ribo-somal sequences. Since the fractionation of DNA on the
basis of G + c content is much lower in csrsOo than in cscl
(Erikson and Szybalski, 1964) it was assumed that ribosomal
sequences would band close to the main peak of single-
stranded DNA. In method (c), however, detection of single-
stranded ribosomal sequences r,rras by filter hybridization as
described above using cDNA made to rRNA as probe.
Y
103.
c RESULTS
1 Isolation of Specific Size Classes of Single-
stranded Chick DNA
(a) Alkaline sucrose eradient centrifuqation
6 Gradients, each loaded with 200 ug ofhigh molecular weight DNA in 0.1 M NaOH, 10 nM EDTA,
rrrere centrifuged as described in Chapter V.8.1. (a)
and fractionated on an ISCO density gradient fraction-ator, model 640. The absorbance at 254 nm r^ras moni-
tored by the optical unit of the ISCO which was coupled
to a W + W recorder. The profile obtained fron all6 gradients were identical and a representative isshown in Figure 5.2. This broad DNA distributionwas subdivided into 5 fractions as shown in Figure
5.2. Corresponding fractions from each gradient were
pooled, neutralîzed, diluted, ethanol precipitated,
re-centrifuged on 5-20% aLkaline sucrose gradients
under the same conditions as before, and fractionated.
The profiles obtained after re-centrifugation are
shown in Figure 5.3. While these profiles demonstrate
that successively higher rnolecular weight fractions
contained DNA of successively larger single-stranded
size, the peaks obtained were very broad indicating
the presence of a wide range of molecular weight
species within each fraction. Accordingly, narror^rer
size ranges hrere selected from these gradients as
shown in Figure 5.3 and the DNA re-centrifuged on
alkaline sucrose gradients under the same conditions
as before. The results from this third alkaline
FIGURE 5.2.
ISOLATION OF SINGLE-STRANDED DNA SIZE CLASSES BY
ALKALINE SUCROSE GRADIENT CENTRIFUGATION 1
High molecular weight DNA (500 ug single-stranded
size >- 50 kb) was loaded on to 5-20% alkaline sucrose grad-
ients and centrifuged as described in Chapter V.8.1. (a).
Gradients were fractionated using an ISC0 density gradient
fractionator coupled to a W and W recorder. The ArtO
profile is shown. The numbered sections at the top of the
figure represent the molecular weight cuts selected for re-
centrifugation.
Sedinentation is from right to left
r-5T4T312-T-lr
0.7
0.6
0.5
0.4
0.3
0,2
0.1
Azst
10 6
intoImls
4
grad ient2 0
Sedimentation
FIGURE 5. 5.
ISOLATION OF SINGLE-STRANDED DNA SIZE CLASSES BY
ALKALINE SUCROSE GRADIENT CENTRIFUGATION 2
MolecuLar weight cuts fron Figure 5.2 were ethanol
precipitated and recentrifuged on 5-20% alkaline suclose
gradients as described in Chapter V.8.1. (a) . Sections
rnarked at the top of each profile represent molecular
weight cuts selected for recentrifugation.
sedirnentation is from right to left. Numbers 1 to 5
correspond to the fraction numbers shown in Figure 5.2,.
f- -l 0.4
0.3
0.2
0'2
0
o1
0
0.05
0.05
04
0r2
Azst
0.1
3 I
4
5
t- -ì
0l- 'r
4grad ient
8mls
6in to
212 '10
Sed imentation
0
FIGURE 5 . 4.
ISOLATION OF SINGLE.STRANDED DNA SIZE CLASSES BY
ALKALINE SUCROSE GRADIENT CENTRIFUGATION 3
Molecular l\Ieight cuts from Figure 5.3 were ethanol
precipitated and recentrifuged on 5-20% alkaline sucrose
gradients as described in Chapter V.8.1. (a). Sections
narked at the top of each profile represent the final mole-
cular weight cuts selected.
sedinentation is from right to left. Numbers 1 to 4
correspond to fraction numbers shown in Figure 5.2.
N.B.: Fraction 5 fron Figure 5.5 contained insufficient DNA
to register aS arL absorbance peak on the 0.2 absorbance
setting on the ISCO.
r-1
0.1
0.0 5
0.0 6
0.04
0.02
0.03
0.02
0.01
0
0.03
0.02
0.01
0
2
Azs¿
0I
3
4 r I
864mls into grad ient
0210
Sed imen tat ion
104 .
sucrose gradient step are shor,vn in Figure 5.4.
Fraction 5 did not contain enough DNA to register as
an absorbance peak on the 0.2 absorbance setting ofthe ISCO. Fractions 1 to 4 showed considerably
narrower peaks than in Figure 5.3 indicating greater
DNA nolecular weight homogeneity within each fraction.Despite this, these peaks \trere stil1 too broad foruse as single-stranded DNA molecular weight size
classes in the density gradient procedure and so a
narror^rer size range was again selected from these
gradients (Figure 5.4) . After concentration by
ethanol precipitation, the single-strand molecular
weight of each of these síze fractions, along with
that of the starting material, was determined by
alkaline agarose ge1 electrophoresis relative to mole-
cular weight markers À phage DNA, 186 DNA and À DNA
digested with restriction endonuclease EcoR, (Figure
s.s) .
From Figure 5.5 it is imnediately obvious
that there was a large amount of DNA breakdown during
the size class isolation procedure. Much of thisbreakdown must have occurred during treatment of the
DNA prior to or during loading on the first of the
alkaline sucrose gradients, since the broad DNA distri-
bution shown in Figure 5.2 contains those regjrons from
which even the lowest molecular weight size class r^ras
derived. Consideration of Figures 5.2 and 5.4 shows
that the position of the cut for a given size fraction
frorn the original gradient (Figure 5.2) is slightly
further toward the bottom of the gradient than the
FIGURE 5.5.
ANALYTICAL ALKALINE AGAROSE GEL ANALYSIS OF SINGLE-
STRANDED DNA SIZE CLASSES OBTAINED FROM ALKALINE
SUCROSE GRADIENT CENTRIFUGATION
1 ug 0f DNA from each of the size cuts shown in Figure
5.4 rtras electrophoresed on a 0.4% alkaline agarose gel as
described in Chapter V.8.1. (b) (ii). I DNA, 186 DNA and
EcoR, digested À DNA (Chapter II.B.9) I^¡ere used as molecular
weight markers. 0, origin.
o
kb46
30
2 0.8
7.2
5.65.154.6
3.3
\R, \ 186 1 2 3 4 lnttl'B1,o
105.
final cuts for the same size fraction (Figure
5.4) indicating that although most of the breakdown
occurred on the first gradient, breakdown continued
to occur on subsequent gradients. Some degree of
breakdown of the DNA could have been tolerated since
the density gradient system required size classes
varying in length between 5 kb and 50 kb. By this
procedure, however, most of the DNA appeared in the
lowest molecular weight fraction (Figure 5.4) which
hras too small to be of practical use (Figure 5.5) .
Increasing the concentration of EDTA in the gradient
and loading solutions failed to reduce the breakdown
problern (data not shown) . Figure 5.5 also shows that
the single-stranded DNA size classes obtained by this
procedure were sti11 heterodisperse with regard to
the molecular weight of their component molecules.
These results, taken together, indicated that the
alkaline sucrose gradient centrifugation procedure
was inadequate for the isolation of relatively homo-
geneous single-stranded DNA size classes over the
range of sizes required for density gradient analysis
of eukaryote gene organization ernploying RNA:DNA hy-
brids.
(b) Preparative alkaline agarose ge1 electro-
phoresis
Agarose gels provide greater resolution
of DNA on the basis of molecular weight than sucrose
gradients. Recently McDonell et al. (I977) have
used alkali as a denaturing solvent for agarose gel
106.
electrophoresis. The possibiLity of using alkaline
agarose gels for the preparative isolation of single-
stranded DNA size classes was, therefore, investigated.
Since the size classes required for the proposed
hybrid-density gradient technique ranged from 5 kb to
50 kb,,low percentage gels (0.5% agarose) which give
an almost linear relationship between 1og MW and
rnobility over this range, were used.
(i) Quality of single-stranded DNA síze
classes
To determine whether preparative
alkaline agarose gels could provide relatively
homogeneous sîze classes of single-stranded DNA,
200 ug of randomly sheared chick DNA in 0.1 M
NaOH, 10 mM EDTA was loaded in the loading slot
of a 20 cm x Z0 cm x 0.5 cfl, horizontal 0.5%
alkaline agarose ge1 and electrophoresed at 100
volts overnight. A slice was taken fron the
side of the gel and stained with ethidium bro-
rnide to determine the position of the DNA. The
DNA containing section of the gel was divided
into 9 slices, each slice was dissolved in SSKI
and the DNA extracted by HAP chromatography as
described in Chapter V.8.1.(c) (i). The DNA
from 6 of these fractions was concentrated by
ethanol precipitation and examined by analytical
alkaline agarose ge1 electrophoresis. The
result of this analysis is shown in Figure 5.6.
The fractions examined showed excellent molecular
FIGURE 5.6.
QUAL ITY OF SINGI,E*STRANDED DNA SIZE CLASSES OBTAINED
FROM PREPARATIVE ALKALINE AGAROSE GELS
0.2 ng 0f randonly sheared chick DNA in 0.1 M NaOH, 10
nM EDTA was electrophoresed in a preparative, horizontaL 0.5%
alkaline agarose gel at 100 volts overnight. The gel was
sliced into 1 cm thick slices taken at right angles to the
direction of electrophoresis. DNA was isolated fron the
agarose slices by solubilization of the agarose in KI followed
by FIAP chromatography (Chapter V.8.1. (c) (i) ) . DNA was
dialysed against T.E. and concentrated by ethanol precipita-
tion. 1 Ug Of DNA fron selected slices hlas electrophoresed
on analytical 0.4% alkaline agarose gels using À DNA, 186
DNA and À DNA digested with EcoR, as molecular üIeight markers.
Numbers represent the distance, in cfl, of the slice
from which the DNA was extracted frorn the origin of the pre-
parative ge1. 0, origin.
kb
46
30
2 0.8
1.25.6
5.154.6
\R, I 2 3 4 6 I 186 \
r07 .
r,.reight homogeneity and ranged in size from about
5 kb to 30 kb.
(ii) Conditions of electrophoresis
To investigate whether alkaline ag-
arose gels could be used on a preparative basis
for the isolation of single-stranded DNA size
classes, 0.5 cn thick 0.5% alkaline agarose gels
were loaded with 600 Ug and 2 mg of randomly
sheared DNA in 0.1 M NaOH, 10 nM EDTA and elec-
trophoresed overnight at 100 volts. The gel
was divided into 1 cm wide strips taken at right-angles to the direction of electrophoresis and
the DNA extracted by centrifugation of the aga-
rose slice (Chapter V.8.1.(c) (iii)). The DNA
from the slices r,tras examined on anaLytical alka-
line agarose gels and the results are shown in
Figure 5.7 . If the preparative gels were over-
loaded with DNA, the DNA extracted from any
given strip would be heterogeneous in molecular
weight. In this situation the DNA from any
given s1ice, when examined on analytical alka-
line agarose gels, would appear as a broad dis-
tribution and there would be a high degree of
overlap between those DNA distributions derived
from different slices. Figure 5.7(a) shows
the results for single-stranded DNA size classes
taken from a preparative gel loaded with 0.6 ng
of DNA, Figure 5.7(b) shows size classes taken
from a ge1 loaded with 2 mg of DNA. The mole-
FIGURE 5.7 .
CAPACITY OF PREPARATIVE AI,KALÏNE AGAROSE GELS
Horizor.ta1- preparative 0.5% alkaline agarose gels
'hrere loaded with 0.6 ng and 2 mg of randomly sheared chick
DNA in 0.1 M NaOH, 10 mM EDTA and electrophoresed at 100
volts overnight. Gels were sliced into 1 cm thick slices
taken at right angles to the direction of electrophoresis.
DNA was extracted by centrifugation of the agarose slice
(Chapter V.8.1. (c) (iii)) and 1 ug from each slice electro-
phoresed on an anaLytical 0.4eo alkaline agarose gel using À
DNA, 186 DNA and À DNA digested with EcoR, as molecular
weight markers.
(a) DNA from a sliced preparative gel loaded with
0.ó ng DNA.
(b) DNA fron a sliced preparative gel loaded with
2.0 rng DNA.
Nunbers represent the distance, in cil, of the slice
from which the DNA was extracted from the origin of the pre-
parative ge1. 0, origin.
kt)46
30
2 0.8
7.2
5.6
kb46
30
2 0.8
b
3456
23456
7 I I 10 11
o
1.25.6
5.154.b
186\R,
:-étl
3
7 I I 10 11
108
cular r^reight distributions for the size classes
obtained with each of these loadings indicated
that the preparative gels were not overloaded in
either case. Accordingly, for all subsequent
experirnents, the initial alkaline agarose ge1
size fractionation was performed on 2 mg of ran-
dornly sheared chick DNA.
To determine whether the rate of the
preparative size fractionation could be increased
2 mg of randonly sheared DNA was electrophoresed
in a preparative alkaline agarose geL at 300
volts for 6 hours. The gel was sliced, DNA
extracted from each slice and examined on anaLy-
tical alkaline agarose gels (Figure 5.8) . The
DNA distributions obtained from the different
slices r^rere broad and distributions from differ-
ent slices showed a high degree of overlap.
This effect was most obvious for the slices
which should have contained the highest single-
stranded DNA molecular weight classes. The
effect was the same regardless of whether verti-
caI or hori zontal gels were used for the initial
size fractionation. For all subsequent
experiments preparative electrophoresis was
carried out at 100 volts overnight.
(iii) Extraction of DNA from a arose slices
2 mg 0f randomly sheared DNA was
electrophoresed in preparative alkaline agarose
gels at 100 volts overnight. The DNA was
FIGURE 5.8.
EFFECT OF VOLTAGE ON THE QUALITY OF SINGLE-
STRANDED DNA SIZE CLASSES OBTAINED FROM PREPARATIVE
AGAROSE GEL ELECTROPHORESIS
Randornly sheared DNA (2 ng) 'hlas electrophoresed on a
horizontaL preparative 0.5% alkaline agarose gel at 300 volts
for 6 hours. The gel was sliced into strips 1 crn wide and
DNA extracted from each slice as for Figure 3.6. DNA (1 ug)
from each slice was electrophoresed on an anaLytical 0.4"ó
alkaline agarose ge1. Numbers correspond to the distance,
in cfl, of the slice from which the DNA was extracted from
the origin of the preparative gel. O, origin.
1 2 3 4 5 6 7 I I 10 11
109.
extracted fron the slices of the gels by three
methods and the recoveries obtained using these
procedures were compared. The amount of DNA
extracted from each slice was determined by
rneasurernent of the A260 of the fractions after
all the manipulations necessary for the appro-
priate isolation procedure had been completed
(Chapter V. B. 2. (c)) . The total recovery from
each gel was then deternined by the surnmation
of the quantities of DNA obtained from each of
the constituent slices. As can be seen from
Table 5.1, centrifugation of the agarose slice
hras the most efficient of the extraction pro-
cedures returning 24% to 33% of the DNA initially
loaded on to the gel. Gel solubíLization with
KI followed by HAP chromatogr:aphy was next,
reproducibly yielding about LZ% of the DNA
initially loaded, while KI gradient centrifuga-
tion yielded 9.9% of the DNA loaded. Under
the conditions of preparative ge1 electrophoresis
used, none of the DNA initially loaded was ever
observed to electrophorese off the bottorn of the
ge1, as determined by ethidiun bromide staining.
The recoveries shown in Table 5.1, therefore,
represent accurate deterninations of the effi-
ciency of recovery of single-stranded DNA from
alkaline agarose gels by each of the three pro-
cedures .
In all subsequent work, 2
was used for the preparative alkaline
rng of DNA
agaro s e
TABLE 5.1.
COMPARISON OF THE EFFICIENCES OF 3 METHODS FOR
EXTRACTION OF SINGLE-STRANDED DNA FROM AGAROSE SLICES
Method Percentage recovery
Agarose solubilization with KIfollowed by HAP chromatography.
Agarose solubilization with KIfollowed by KI density gradientcentri fugati on .
Centrifugation of the agaroses lice .
10-12.5
9.9
24- 33
2 mg Of DNA was electrophoresed on a preparative hori-
zontal- alkaline agarose ge1 overnight at 100 volts. The ge1
was sliced into 1 cm strips at right angles to the direction
of electrophoresis and all strips fron any given gel were
subjected to the same extraction procedure (Chapter V.8.1. (c)).
The amount of DNA extracted from each slice was estimated by
reading the ArU' of each resultant fraction after the frac-
tion had been subjected to all the steps necessary for the
particular extraction procedure. The amount of DNA extracted
from each slice of a given gel hlas then sumrned giving the
total amount of DNA extracted fron the gel. This value was
divided by the total amount of DNA loaded on the gel and
nultiplied by 100 to give percentage recovery.
N.B.: Under the conditions of preparative electrophoresis
employed, ro DNA rvas ever observed to run off the end of
the ge1 as deternined from ethidiun bronide staining.
110 .
ge1. Electrophoresis was performed at 100
volts overnight and DNA was extracted from the
gel slices by centrifugation of the slice.
2 Studies on the Hybrid-densit y Gradient Procedure
for the Analysis of Gene Arran gement
(a) Detection of T00% hybrid and single-stranded
ribosomal sequence densities
The first requirernent was to determine the
densities of 100% hybrid and single-stranded DNA.
L00% Hybrid was generated as described in Chapter V.8.3
and loaded on to CsrS0O gradients in pre-hybridization
buffer. In a sinilar manner, 100 ug of DNA was alkali
denatured, neutralized and loaded on to CsTSOO grad-
ients. After centrifugation, the gradients were
fractionated from the botton. The banding position
for pure hybrid (Figure 5.9 (a) ) was deternined by TCA
precipitation of each fraction followed by the deter-
mination of the amount of radioactivity in the pre-
cipitates. The banding position of single-stranded
DNA was determined by reading the A,UO of each fraction(Figure 5.9 (b) ) . The banding position of L00% hybrid
varied in different experiments between 1.548 g/cc
and 1.552 g/cc. This variation was the same regard-
less of whether the hybrid was generated using keratin
nRNA:cDNA hybrids or rRNA:cDNA hybrids. The density
of 100% hybrid was therefore taken to be 1.55 E/cc
under these conditions. The density of single-
stranded DNA ranged from 1.452 to 1.456 g/cc in
independent experiments and so the actual density for
FIGURE 5.9.
DENSITY OF SINGLE-:STRANDED DNA AND 1OO% HY-BRID IN
CsSO GRADIENTS4
L00eo Hybrid was prepared from 32p 1abe11ed keratin
cDNA and keratin nRNA as described in Chapter V.8.5.
100,000 c.p.m. Of this hybrid was added to 2 mL of pre-
hybridi zation buffer then added to a solution of CsrSOO to
give a final volurne of 9 nl and a density of 1.5 E/cc (Chap-
ter V.8.4). 100 ug 0f high molecular weight chick DNA was
denatured with alkali, neutralized, added to 2 mI of pre-
hybridization buffer and the gradient prepared as described
for I00% hybrid. Conditions of centrifugation were as des-
cribed in Chapter V.B.4. Gradients hlere fractionated from
the bottom and densities determined by refractometry. Hy-
brid was detected by TCA precipitation of the fraction con-
tents followed by radioactive counting. Single-stranded
DNA was detected by reading the ÃZOO of the fractions.
(a) I00% hybrid.
(b) Single-stranded DNA.
Density increases from right to left. Densities of
I0\eo hybrid and single-stranded chick DNA are indicated.
fiÌIo^Fb
x
=o-c)4o-
N(Y:t
a
10
I
2.0
1.6
1.2
0.9
o/cc1.548
I
'!0 '¡ 5
Fract ion Numbe r
2
o(os¡
0.4
05 25
Top
10 155 2b 201
Bottom20
111 .
single-stranded DNA under these conditions was taken
to be 1.454 g/cc.
(b) Pre-hyb ridization and detection of hybrid
in CsrSO, gradients
Having determined the liniting density
values it was necessary to be able to detect the hy-
brid generated in the pre-hybridization step after
fractionation in the CsrS0O gradients. Pre-hybridiz-
ation was performed using L25r labelled 18s and 28s
rRNA and 100 ug of DNA of single-stranded length >,50
kb. Hybrids were centrifuged in CsrS0O gradients
and the gradients fractionated from the botton.
Gradient fractions were divided into two. The
fractions of one of these sets were treated with pan-
creatic RNA'ase followed by TCA precipitation (Chapter
V.8.5.(a)) while the fractions of the other set were
filtered through nitrocellulose discs to selectively
inrnob iLize the iodinated hybrid (Chapter V.8.5. (b) ).
The results from this experiment are shown in Figure
5.10(a). The figure shows that, after RNArase treat-
ment, much of the RNA was still in TCA precipitable
form. If the ribosomal sequences represent 0.035%
of the chick genome (see Chapter IV.A.) and the spe-
cific activity of the L25r labelled RNA used in the
pre-hybridization step was 2 x 105 c.p.n./ug for each
species, then the maximum number of counts expected
in the hybrid generated under these conditions, is
about 1.75 x 105 c.P.fl. , taking asymmetric transcrip-
tion into account. The large arnount of undigested
FIGURE 5.10.
(a)
(b)
HYBRTD DETECTION TECHNIQUES
Hybrids were prepared using 100 ug of high molecular
weight DNA (> 50 kb) and 20 ug each of L25r labelled
18S and 28S rRNA (Sp.Act. 5 x 105 c.p.m./ug) as des-
cribed in Chapter V.8.2. The pre-hybridization solu-
tion was loaded on to CsrSOO gradients and centrifuged
as described in Chapter V.8.4. Gradients were frac-
tionated from the bottorn, densities deternined by re-
fractometry and the fractions divided in two. One set
of fractions ltlas treated with pancreatic RNArase as
described in Chapter V.8.5.(a), the other set being
filtered through nitrocellulose discs as described in
Chapter V.8.5.(b). AZ6O (single-stranded chick DNA),
I RNA! ase treatment, tr filter binding.
The density of single-stranded chick DNA is shown.
Hybrids were prepared using 50 ug of high molecular
weight DNA and 20 ug each of unlabelled 18S and 28S
rRNA. Gradients were prepared and centrifuged as
above. After fractionation of the gradients fron the
bottorn, densities were determined by refractometry, DNA fron
individual fractions I^Ias immobilized on nitrocellulose
discs and the filters challenged with L25I labeIled 185
and 28S rRNA (Sp.Act. 2.5 x L07 c.p.n./ug) as described
in Chapter IV.B.3. AZOO (single-stranded chick
DNA), r I25I c.p.m. hybridized (Hybrid).
Densities for single-stranded chick DNA and hybrid are
indicated.
t.4 5 4 s/cc
I
25
Top
80
60
120
100
(\¡to-14
NIo
Io
x
ECLI
rlt(\¡
40
20
EÊ(,,
r¡)ô¡
otON
10
I6
4
2
îtQ'
,!3.c¡
Eèo
útN
80
60
40
20
0'3
0.2
0.1
0.4
f12
10
Fraction15
Number5
Botto m
2510 15
1.456 s/cc1.493 sf cc
205b
II
20
TLz.
iodinated RNA present in the bottorn half of the grad-
ient would, therefore, nake detection of a hybrid
peak, which could span a number of fractions, quite
difficult. Backgrounds were much lower when hybrid
retention on nitrocellulose discs 'hias enployed
(Figure 5.10 (a) ) . There was, however, îo obvious
hybrid peak. In addition, the number of filter
retained counts was higher at the top of the gradient
than elsewhere. This rnay be due to non-specific
entrapment of r25r labelled rRNA with single-stranded
DNA since single-stranded DNA will bind to nitrocellu-
lose. ThiS cannot be the whole explanation, however,
since the filter retained counts profile does not
fo11ow the AZOO profile. These results l4lere shown
to be reproducible even when whole fractions were sub-
jected to either of these treatments.
Owing to the failure of these simpler
techniques to a11ow the detection of hybrids in csrsoo
gradients, a modification of the filter hybridization
proceduïe,used by Birnstiel et al. (1968) to study the
arrangement of ribosomal sequences in Xenopus laevis,
r,rras ernployed (Chapter V.8.5.(c)). 20 ug Each of
unlabelled 28S and 18S rRNA was used in the pre-
hybridi zation reaction with 50 ug of DNA of size >/ 50
kb. Hybrids were loaded on to csrsoo gradients and,
after centrifugation, the gradients were f.ractionated
from the bottom. The DNA content of the fractions
was determined by measurement of their AZOO, then the
DNA was irnnobiLized on nitrocellul0se filters.
Filters were challenged with L25I 1abelled rRNA
113 .
6(specific activity 25 x 10 c.p.m./ug), the filterswashed and the radioactivity bound to the filters
deternined. The results are shown in Figure 5.10(b).
Single-stranded DNA exhibited a density of 1.456 g/cc
while the hybrid appeared as a sharp peak at a density
of 1.50 E/cc. Thus, by using filter hybridization
it was possible to detect the position of the hybrid
in CsTSOO gradients.
To ensure that the peak ascribed to hybrid
in Figure 5.10(b) was in fact due to hybrid and not
some odd banding behaviour of single-stranded ribo-
somal sequences in CsrS0O, single-stranded DNA of
length > 50 kb was subjected to pre-hybridization con-
ditions (without RIIIA) followed by CsrSOO centrifugation. After
centrifugation, the gradient was fractionated, the
DNA distribution deternined by measurement of the 4260,
the DNA fron each fraction inmobiTized on nitrocellu-
lose filters, filters challenged with L25r 1abelled
rRNA and the distribution of radioactivity compared
with that in Figure 5.10(b). Figure 5.11 shows that
single-stranded ribosomal sequences banded to the
heavy side of the Ar60 profile for single-stranded
chick DNA, a result not initially expected. It was
not possible, however, to completely explain the
resolution between the hybrid and single-stranded DNA
peaks of Figure 5.10(b) on the basis of the G + C con-
tent of the ribosornal sequences. This suggested that
the filter hybridization technique was in fact detec-
ting the generated hYbrid.
LL4.
From the results shown in Figure 5.10, ithras apparent that the best nethod for detection ofhybrids in the CsTSOO gradients was filter hybridiza-
tion. Since the density of single-stranded ribosonal
sequences did not coincide with that of total DNA
(Figure 5.11) and since the average G + C content of
these sequences may vary with DNA fragnent size, ithias necessary to determine the density of these
sequences for each size class of DNA used. These
two requirenents were most easily satisfied by immo-
bilizing the DNA fron each fr:action on nitrocellulosediscs, splitting the discs in two and challenging one
set of half filters with L25T 1abe1led rRNA and the
other set with cDNA made to rRNA. Clearly the use
of cDNA hybridization to indicate the density of the
single-stranded sequence under study assumes that all
these genes are transcribed fron the same strand of
the DNA. Thus the importance of control experiments
subjecting single-stranded DNA to pre-hybridization
conditions in the absence of any RNA, followed by
CsTSOO gradient analysis, cannot be overstressed (see
DrscussroN).
(c) Comments on pre-hybridization conditions
The hybrid-density gradient rnethod for the
analysis of gene arrangement relies rnainly on two con-
ditions being net during the pre-hybridizatíon reac-
tion: -
(i) RNA:DNA hybridization must be driven
FIGURE 5.11.
DENS ITY OF SINGLE-STRANDED RIBOSOMAL SEQUENCES
High nolecular weight chj-ck DNA (50pg) was denatured with
alka1i, neutralized, added to pre-hybridization buffer and
the CsTSOO gradient prepared and centrifuged as described in
Chapter V.8.4. The gradient was fractionated from the bot-
torn into 0.3 ml fractions, densities determined by refracto-
metry and the AZOO of each fraction deterrnined. DM from each
fraction was immobiLized on nitrocellulose discs and the
filters challenged with L25I labelled 185 and 285 rRNA (Sp.
Act. 2.5 x LO7 c.p.m. /uÐ to detect ribosomal sequences
(Chapter IV.B.3) .
AZOO (single-stranded chick DNA).
L25I c.p.rn. hybri ð,izeð, (single-stranded ribo-
somal sequences).
Densities of single-stranded total chick DNA and ribo-
sornal sequences are indicated.
I
NIo!F
1'46
I
20
t slcc
1.453 e /cc
I
o(gN
6
4
x
T'o
.T
=L.cr
ECL(,
H
1.5
1.0
0.5
0
f'2C\IF
0
Bot to m10
Fract ion15
Number5 25
Top
115.
ahead of DNA reassociation.
(ii) The conditions of pre-hybridization
must not lead to the breakdown of
single-stranded DNA.
It is known that the rate of DNA reasso-
ciation increases with increasing length of the reac-
ting molecules and decreases with increasing viscosity
of the solution. Taking these factors into account,
it was possible that the apparent extent of reassocia-
tion attained, under the conditions of pre-hybridiza-
tion employed, rnight have been greater than that
expected for molecules of average length 500 bases at
arL equivalent value of Cot, If this was so, then
reassociation of the sequences under study would not
pernit saturation of the DNA with RNA. In addition,
given that RNA saturation of the DNA occurred,
reassociation of the DNA surrounding the sequences of
interest would lead to the production of double-
stranded complexes which could carry additional single-
stranded DNA into the hybrid structure. 0n density
gradient analysis, both of these effects would be
expected to produce a broad hybrid peak of reduced
density with a shoulder tailing towards the light side.
In a sirniLar way, the shape of the single-stranded
peak, as determined by cDNA hybridization, would be
expected to be of higher density and broad with a
shoulder tailing towards the heavy side.
To determine whether DNA reassociation
11ó .
was affecting the density gradient analysis of the
ribosornal cistrons, pre-hybridizations were perforrned
using 50 ug of high molecular weight DNA and two dif-ferent levels of 18S and 28S rRNA. The resultanthybrid structures hrere analysed on CsTSOO gradients
and a qualitative assessment of reassociation effectswas made on the basis of the respective shapes of the
peaks corresponding to the hybrid and single-stranded
ribosomal sequences.
Figure 5.I2(a) shows the profiles ob-
tained using 20 ug each of 18S and 28S rRNA in the
pre-hybridization reaction. The hybrid peak was
broad and birnodal with the smaller peak coinciding
with the rnain peak of cDNA hybridization. The cDNA
hybridization pattern consisted of a sharp peak at a
density similar to that expected for single-stranded
ribosomal sequences (Figure 5.11) with a shoulder
tailing off to the heavy side. This pattern was
compatible with that expected if some reannealing ofsequences within the ribosomal cistrons had occurred
during the pre-hybridi zation reaction.
Figure 5.12 (b) shows the profiles obtained
using 60 ug each of 18S and 28S rRNA in the pre-hybrid-
ization reaction. The L25r 1abe11ed rRNA hybridiza-
tion pattern indicated that the hybrid peak was stillbroad but was no longer birnodal. The cDNA hybridiza-
tion showed a main peak of hybridization similar to
that for single-stranded ribosornal sequences and a
trailing heavy shoulder. It should be noted, however,
FI GURE 5 .T2 .
AMOUNT OF RNA REQUIRED FOR RNA:DNA HYBRIDIZATION
IN THE ABSENCE OF DNA REANNEALING
Hybrids were generated using 50 ug of high molecular
weight chick DNA and 20 ug or 60 ug each of 18S and 28S rRNA
as described in Chapter V.8.2. The pre-hybridization solu-
tion was added to CsTSOO solution and the gradients prepared
and centrifuged as described in Chapter V.8.4. Gradients
r^rere fractionated from the botton, densities determined by
refractometry, the DNA fron each fraction inmobilized on
nitrocellulose filter discs and these filters split in half.
One set of half filters r^ras challenged with L25r labelled
rRNA (Sp.Act. 3 x I07 c.p.m./ug) to detect ribosomal sequences
present in the hybrid and the other set was challenged with
cDNA made to rRNA to determine the density of single-stranded
ribosomal sequences.
(a) Hybrids generated using 20 ug each of 18S and
2 85 rRNA; r T25I c .p. m. hybri d.ized (hybrid) ,
tr 3H c.p.m. hybridized (single-stranded ribo-
somal sequences).
(b) Hybrids generated using 60 ug each of 18S and
28S rRNA; r L25I c.p.m. hybridized (hybrid),
tr 5H c.p.m. hybridized (single-stranded
ribosomal sequences).
Densities of hybrid and single-stranded ribosonal sequences
are indicated.
1'ae9 s/c
1.468 s/ccI
c 30
25
15
10
25
20
î5
t0
a
10
I
6
4
2
25
20
15
5
10
Fraction15
Number
x20x
I
o
Eo.!E
.ct
EctC¡
Id:t
Ie
x
E'o.I:¡t
EcloI
câ
10
!o.!
=¡t
Eçlo
¡f)(\¡
Io
.9
tlt
.!:¡l
Eo(,
rtl¡
5
0
Io
x
5
05
t0 155b
.251."
I
20 t.tt o s1'484sfcc
I
Top2520
Bottorn
LI7 .
that this heavy shoulder was not as marked as that
shown in Figure 5.LZ(a). Under these latter condi-
tions, therefore, it appeared that DNA reassociation
within the ribosonal cistrons was less narked.
It can be seen from Figure 5.IZ that the
density of the peak of cDNA hybridization in these
experinents did not vary significantly from that for
single-stranded ribosornal sequences shown in Figure
5.11. From this observation and the lact that the
difference between hybrid peaks and shoulders rtrere
clearly discernible, it was concluded that DNA reasso-
ciation had only affected the density of a proportion
of the molecules present (about 25% in Figure 5.12(b))
and that the peak of LzSr 1abel1ed rRNA hybridi zation
and cDNA hybridization probab1-y represented the true
densities of saturated hybrid and single-stranded
ribosomal sequences respectively. It was also
apparent that increasing the amount of RNA used in
the pre-hybridization step, accompanied by a shorter
incubation time, or decreasing the amount of DNA used
to form the hybrid, should further decrease the effect
of DNA reassociation on density analysis.
To determine whether single-stranded DNA
size classes hrere degraded during the pre -hybridLza-
tion process, samples from various size classes of
single-stranded DNA ltlere divided in two with one half
being subjected to denaturation, neutralization and
pre-hybridization conditions while the other half was
used as a control. Figure 5.13 shows the comparison
FIGURE 5.15.
THE EFFECT ON THE CONDITIONS OF PRE-HYBRIDIZATION
ON SINGLE-STRANDED DNA SIZL,
Four DNA size classes l^iere isolated from slices of a
preparative alkaline agarose gel as described in Chapter V.
8.1. (c) (iii). 20 ug 0f each size class I^Ias subjected to
pre-hybridization conditions in the absence of RNA (alkali
denaturation, neutralization, made up to 2 mL of 2 x SSC,
incubated at 60oC for 30 min). I ug Of each of these was
ethanol precipitated, redissolved in 40 u1 of 0.1 M NaOH, 1
nM EDTA and electrophoresed on an analytical 0.4% alkaline
agarose ge1 as described in Chapter V.8.1. (b) (ii). 1 ug
0f DNA from the same initial síze class \tlas electrophoresed
in the adjacent track. À DNA, 186 DNA and À DNA digested
with EcoR, urere used as nolecular weight narkers. Numbers
represent the distance, in cilr of the slice from which the
DNA was extracted frorn the origin of the preparative gel.
0, origin.
o
kb46
30
2 0.8
.2
.615.6
7
55.4
3.3
113344before af ter l¡efore af ter before af ter
55before after
RÀ 186
*SF¿;
"r*.,þl
118 .
of the size of these single-stranded size classes
before and after pre-hybridization conditions as deter-
nined by alkaline agarose ge1 electrophoresis. These
results indicated that no appreciable DNA breakdown
occurred during pre-hybridization.
(d) The relationship between density and the
proportion of the DNA nolecule in hybrid
form
To investigate the relationship between
the proportion of the DNA molecule in hybrid form and
the density of this hybrid, relative to the densities
of the two extremes of single-stranded DNA and 100%
hybrid, the ribosomal cistrons of chick were used as
the model system. Restriction analysis of total
chick DNA and cloned ribosomal sequences has shown
that the ribosomal repeat unit is about 27 kb long
and contains I copy each of the structural genes for
18S and 28S rRNA (McClements and Skalka, L977).
60 ug 0f 28S rRNA was hybridized to Z0 ug of single-
stranded DNA from different size classes under the
conditions outlined in Chapter V.8.2. The resultant
mixture was centrifuged in CsTSOO gradients, the grad-
ients fractionated and fractions treated as in Chap-
ter V.8.5. (c). Since only 28S rRNA was used to form
the hybrid, only r25r labelled 28s rRNA was used as
probe to detect it. The profiles obtained using
single-stranded DNA size classes ranging in size from
5.9 kb to L4.6 kb are shown in Figure 5.14. The
sharp symnetrical peaks of cDNA hybridizatron,
FIGURE 5.14.
HYBRID-DENSITY ANALYSIS OF CHICK RIBOSOMAL CISTRONS
USING SINGI,E-STRANDED DNA OF DIFFERENT SIZES
Single-stranded DNA size classes r{Iere isolated fron pre-
parative alkaline agarose gels as described in Chapter V.8.1.
(c) (iii). 20 ug Of DNA from each size class and 60 ug of
285 rRNA was used in the pre-hybridization reactions. Pre-
hybridization mixes were centrifuged in CsTSOO gradients
(Chapter V.8.4) and the gradients fractionated from the bot-
tom (0.3 ml fractions). Densities were determined by re-
fractometry. DNA from each fraction was innobíIized on
nitrocellulose filter discs, filters split in half, one set
of half filters being challenged with 125I labelled 285 rRNA
(Sp.Act. 3 x L07 c.p.m./ug) to detect the hybrid and the
other with 3H cDNA to rRNA to detect the density of single-
stranded ribosomal sequences.
(a)
(b)
(c)
(d)
(e)
DNA L4.6 kb; I 1'25r c.p.n. (hybrid), E 3H
c.p.m. (single-stranded ribosonal sequences) .
DNA 1 2 kb; r L25I c.p.m. (hybrid), E 3H
c.p.m. (single-stranded ribosomal sequences) .
DNA 9 kb; r L257 c.p.m. (hybrid), o 5u
c.p.m. (single-stranded ribosonal sequences) .
DNA 7 .2 kb; r rzsr c.p.m. (hybrid) , tr 3u
c.p.rn. (single-stranded ribosomal sequences) .
DNA 5.9 kb; r Tzsr c.p.m. (hybrid) , E 5n
c.p.m. (single-stranded ribosonal sequences) .
Densities of hybrid and single-stranded ribosomal sequences
are indicated.
a
1.4685 s,cc
I
1'508 sf cc
I
b t.5l
I6g fcc
l'4685 sfcc
d l'a0lcrcc1'525s1""
8Sgfcc t'1?
l0
I6
I2
14
12
10
I
6
I
2
t0
e
6
a
2
10
t6
ñIo
x
!to.!€¡tà
ECIt)
-
l0
t6
a
t0
a
2
a
2
t0
NIC'
x
!o.!P¡¡àE
ECLu
6d
E
6
a
2
2
25
Top
t0
E
0
I2
IE
60
I
2
4
2
l0 15
Fract ion N um be r
20
Bottom
119
denoting the density of single-stranded ribosomal
sequences, suggested that 1itt1e or no DNA reassocia-
tion affected these sequences under the conditions of
pre-hybridizatíon used. The hybrid peaks, however,
were very broad and it was difficult to obtain accu-
rate estimates of the density of the hybrid. The
hybrid densities shown in Figure 5.14 'hlere determined
by reading off the density corresponding to the nid-
point of. a line drawn horizontally across the hybrid
peak at half the peak height.
Knowing the size of the single-stranded
DNA used in hybrid formation, and knowing that 28S
rRNA is 5 kb long, it was possible to calculate the
average proportion of any given hybrid molecule
expected to be in hybrid forrn by dividing the length
of 28S rRNA by the length of the single-stranded DNA
size class. This will only hold for single-stranded
DNA up to the length of one repeat unit. The experi-
mental value for the proportion of the DNA molecule
in hybrid form was calculated by dividing the density
difference between the generated hybrid and single-
stranded ribosomal sequence by that between I00%
hybrid and single-stranded sequences. Multiplication
of this value by 100 gives the percent saturation.
A comparison of theoretical and experimental estimates
of the percent of the DNA saturated with RNA for the
DNA size classes used in Figure 5.14 is shown in
Table 5.2. The data in Table 5.2 indicate that there
was very 1ittle sinilarity between the figures
obtained by these two independent estirnation procedures.
TABLE 5,2
THEORETTCAL AND EXPERIMENTAL PERCENT SATURATION VALUES
FOR THE DNA SIZE CLASSES OF FIGURE 5.14
Profi 1efrom Fig.5.14
Theoreti calPercent saturation
ExperinentalPercent saturation
Size(kb)
abcde
T4.6T2
34.24r .755.669 .484 .7
48.558.353.070 .97? L
97?5.9
Single-stranded DNA size classes were prepared by pre-
parative alkaline agarose ge1 electrophoresis. Single-
stranded DNA sizes were calculated by anaTytical alkaline
agarose gel electrophoresis using À phage DNA, 186 phage
DNA and À DNA digested with restriction endonuclease EcoR,
as molecular weight markers. Pre-hybridization and
density gradient centrifugation conditions were as described
in the legend to Figure 5.14. Theoretical percent satura-
tion values were calculated by dividing the length of 285
rRNA by the length of the DNA size class being used and
multiplying by 100.
Experirnental percent saturation was calculated by
dividing the density difference between generated hybrid
and single-stranded ribosomal sequences by that between
100% hybrid and single-stranded sequence and multiplying by
100.
I20.
This effect was expected for the hybrids produced
using the lower molecular weight DNA fragments (5 kb
to 8 kb) since many of the hybrid molecules produced
under these conditions would be expected to have
single-stranded RNA sections producing abnormally
high hybrid densities and hence high percent satura-
tion estinates. Where higher molecular weight DNA
(9 kb - 50 kb) was used in the pre-hybridization,
however, a much closer approxination of experimental
results to theoretical estimates was expected since
strucutres containing single-stranded RNA would be
expected to be far less frequent. An irnproved agree-
ment, however, \^Ias only observed when high molecular
weight DNA was used in the pre-hybridization reaction.
These results suggested that there was not
a sirnple relationship between the theoretically
estinated and the experimentaLly calculated percentage
of the DNA saturated with RNA. Accordin9ly, theore-
tical and experinental percent saturations were deter-
mined for more DNA size classes, the data pooled and
the experimentally determined percent saturation plot-
ted against that expected for 285 rRNA hybridized to
the different lengths of single-stranded DNA. These
results are shown in Figure 5.15. The scatter of
points is sufficient to discourage aîy attenpt to
draw a line of best fit through them. This result
suggests that while there was a trend for the hybrid
density to decrease with increasing DNA length, the
basic information required for the calculation of the
FTGURE 5.15.
RE LATIONSHIP BETIlTEN THEORETICALLY AND EXPERIMENTALLY
DERIVE D ESTIMATES FOR THE AVERAGE PERCENTAGE OF A DNA
MOLECULE CARRY ING RIBOSOMAL CISTRONS TN HYBRID FORM
Hybrid-density analyses were perforrned for a number
of single-stranded DNA size classes as described for Figure
5.14. From this experimental data an experinentally
derived estimate for the average percentage of a DNA nole-
cu1e, containing ribosomal cistrons, in hybrid form (per-
cent saturation) could be made as described in the legend to
Table 5.2. Knowing the length of the single-stranded DNA
and the length of 28S rRNA, the theoretical value for the
percent saturation could be deternined as described in the
legend to Table 5.2. The figure shows the retrationship
between these two values for a number of different single-
stranded DNA size classes.
N. B. : All DNA size classes used l\Iere less than 1 chick ribo-
somal repeat unit in length.
90
80
70
a
o
10 20
a
30 40
Theoretical
o
50 60
o/o Satu rat iolr
o
70 80 90
o
o
60
50
40
30
20
C
.9
r!
rn
Øô
è-
;o
E
ox
t¡¡
a oa
o
o
o o
o
10
0
I2T.
experimentally derived percentage saturation was
insufficiently precise to perrnit any meaningful com-
parison with those values calculated by dividing the
length of the saturating RNA by the molecular weight
of the DNA size class being used. There are many
possible sources of error in the experimental data
obtained, and these will be dealt with in detail in
the Discussion.
For the hybrid-density gradient procedure
to be of any value in the elucidation of gene arrange-
ment it was essential that some discernable relation-
ship exist between the proportion of the DNA nolecule
in hybrid form and the density of that hybrid relative
to single-stranded DNA and 100% hybrid. The inaccu-
racies of the density gradient procedure, however,
render the technique inadequate for this purpose.
For this reason no attempt was made to apply this
systern to the question of keratin gene organization.
D. DISCUSSION
I Isolation of Sinsle-stranded DNA Size Classes
Figures 5.2 to 5.4 show the successive stages of
purification of single-stranded s:-ze classes of DNA obtained
by repeated cycles of alkaline sucrose gradient centrifuga-
tion. It is obvious from these profiles that the procedure
does not a11ow the selection of particularly homogeneous
size classes of single-stranded DNA. This is reinforced
by the anal-yticaL alkaline agarose gel patterns obtained
from the purified size classes (Figure 5.5). The data in
L2 2.
Figure 5.5 also suggest that a large amount of DNA break-
down occurs during the alkaline sucrose gradient procedure.
After the final alkaline sucrose selection step, the frac-
tion containing the most DNA is that containing the smallest
molecular weight class (< 3 kb). This DNA is too small to
be of any practical use in the proposed hybrid-density
gradient procedure for investigating gene arrangement.
Since the size classes were heterogeneous and most of the
DNA was too degraded, the alkaline sucrose gradient rnethod
for selection of single-stranded DNA size classes was not
considered any further.
From Figure 5.6 it is apparent that preparative
alkaline agarose ge1 electrophoresis can provide single-
stranded DNA size classes of sufficient hornogeneity for the
proposed hybrid-density gradient analysis. Figures 5.7 and
5.8 show that as much as 2 mg of randonly sheared DNA can be
fractionated per run on a preparative 0.5% alkaline agarose
gel of dimensions Z0 cm x 20 cm x 0.5 cfl, if electrophoresis
is carried out at 100 volts overnight. The rnajor limitation
of the preparative alkaline agarose gel nethod lies in the
extraction of the DNA from agarose (Table 5.1) .
Three rnethods for extracting DNA from agarose
rltrere tried; solubilization of agarose with KI followed by
HAP chromatography, KI gradient centrifugation and centri-
fugation of the agarose slice followed by collection of the
Supernatant . The last of these I\Ias the most Success ful ,
yielding up to 30% of the DNA initially loaded on the gel.
The success of this centrifugal squeezing procedure in
extracting DNA from agarose, would be expected to decrease
r23 .
as the percentage of agarose in the gel increased. Logically,
the methods involving solubi1-ization of agarose, followed by
separation of DNA and agarose by physical procedures, should
have been the nost successful extraction techniques. The
most likely reason for their lack of success in the present
instance is that the DNA was being handled in its single-
stranded form. Since both solubiLization DNA extraction
procedures involved extensive dialysis at some point (Chap-
ter V. B. 1 . (c) ) most of the losses us ing these methods l^Iere
probably incurred by the single-stranded DNA adsorbing to
the dialysis tubing.
To sunmarize, the results frorn the first part
of this chapter suggest that single-stranded DNA size classes,
suitable for use in the proposed hybrid-density gradient
nethod for analysing gene organization, can be obtained by
preparative alkaline agarose gel electrophoresis of randomly
sheared chick DNA. After sli-cing the gel at right angles
to the direction of electrophoresis, the DNA from each
slice can be extracted by centrifugation of the slice and
collection of the supernatant. All single-stranded DNA
size classes used in the experiments described in the second
part of this chapter hiere obtained using this procedure.
Studies on the Hybrid-densitY Gradient Procedure2
for the Analysis of Gene Arrangement
When these studies were initiated, the techniques
of restriction enzyme analysis, Southern transfers (Southern,
1975(b)) and DNA cloning (wensink et ãI., L974) were not in
general use. There had been, however, a number of highly
infornative studies on the arrangement of the sequences for
124.
18S and 285 rRNA in Xenopus laevis (Wallace and Birnstiel,
1966; Birnstiel e! al., 1968 and Brown and Weber, 1968b), in
numerous other eukaryotes (Sinclair and Brown, 1971) and
the arrangement of tRNA genes of Xe4qpqq laevis , aIL of
which exploited the density difference between RNA:DNA
hybrids and single-stranded DNA in CsC1. Investigations
designed to ascertain the value of a nodification of these
techniques for deterrnining the arrangement of tandenly
linked fanilies of eukaryotic genes were therefore begun.
The ribosomal cistrons of chick were chosen as the experi-
mental system because of the ready availability of large
amounts of highly purified chick 18S and 28S rRNAs. If
the procedure proved to be successful it was planned to use
it for the analysis of the arrangement of keratin genes in
the chick genome. During the course of this work the
arrangement of the chick ribosomal cistrons, as determined
by restriction analysis of total chick DNA and cloned chick
ribosomal sequences using the Southern transfer technique,
r^ias published by McClements and Skalka G977 ) allowing a
critical evaluation of the hybrid-density technique to be
made. The project was persued, nevertheless, in the belief
that it might provide valuable information about average
gene arrangements. These results could then be compared
with those obtained from Southern transfer analysis, pernit-
ting more obj ective interpretations of the gene arrangement
to be made.
Requirernents for the hybrid-density
gradient procedure
(a)
For the technique to be useful for the
125.
analysis of the average arrangement of a tandenly
repeated fanily of genes, it was necessary to deter-
mine the relationship between the proportion of a
single-stranded DNA molecule in hybrid form and the
density of the hybrid structure relative to those of
the two extremes of single-stranded DNA and L00%
hybrid. Before this relationship could be investi-
gated, however, it was necessary to determine the
lirniting densities and to be able to detect the hybrid,
generated in the pre-hybridization reaction, in the
density f.ractionating nedium. CsrS0O was selected
as the fractionating rnediurn rather than CsCl since,
in earLy studies, 100% hybrid was found to pellet at
the bottorn of the latter type of gradient (results
not shown). Figure 5.9 shows that it was possible
to obtain good separation of L00eo hybrid from single-
stranded chick DNA under the conditions of centrifu-
gation chosen (density 1.5 E/cc, 28r000 r.P.m., Ti50
rotoï, 20"C, 60-70 hours). The peak obtained for
100% hybrid was always quite broad which made the
accurate estination of this lirniting density difficult.
The breadth of this peak was presumably due to the
diffusion of the smal1 hybrid molecules around their
actual banding density. so as to reduce the subjec-
tivity involved in estirnating densities from broad
peaks, a line was drawn across the peak at half the
peak height and the rnid-point of this line was taken
as being the actual banding position of the structure.
This technique was used for all density estimates
reported in this chaPter.
L26.
Three nethods for detecting the hybrid
generated in the pre-hybridization reaction r^rere
attenpted. The first involved the use of ]-25I
labelled rRNA to form the hybrid, fractionation of the
hybrid on CsTSOO gradients followed by digestion of
the unhybridi r"d r25r labelled RNA in each fraction
with pancreatic RNA'ase A and TCA þrecipitation of the
resistant material (Figure 5.10 (a) ) . Much of the RNA
pelleted as would be expected from its density
(SzybaIski, 1968), but rnuch of it was also resistant
to RNA'ase under the conditions of digestion used,
causing high backgrounds of radioactivity especially
in the r.ractions fron the botton half of the gradient.
This resistant L25I 1abe11ed material was presumably
28S rRNA since it has a high degree of secondary
structure. Sinilar problems with the digestion ofT25r labelled 28s rRNA with pancreatic RNA'ase A have
been encountered by another rnember of this Department
(4. Robins, personal communication).
It was reported by Wallace and Birnstiel
(1966) that hybrids made with radioactively labelled
RNA bound to nitrocellulose filters while unhybridized
RNA did not. When hybrids were generated with L25I
labelled rRNA and the fractions from the csrsoo grad-
ients filtered through nitrocellulose under these con-
ditions, ûo unhybri dized RNA bound to the nitrocellu-
1ose, but there was no discernable hybrid peak either
(Figure s.10 (a) ) .
The third method was a slight modification
L27 .
of that used by Birnstiel et a1. (1968) in their study
of the ribosomal cistrons of Xenopus laevis. Pre-
hybridization hlas carried out using unlabelled rRNA,
the nixture fractionated on CsZSO+ gradients, the DNA
from the resulting fractions was inmobi1-ized on nitro-
ce11u1ose filters and the filters challenged with125I labelled rRNA. Figure 5.10(b) shows that it was
possible to detect the hybrid using this technique.
When single-stranded DNA was subjected to pre-hybrid-
ízation conditions in the absence of RNA, centrifuged
in CsTSOO gradients and the banding position of ribo-
somal sequences determined by filter hybridization,
these sequences were found to be more dense than the
bulk of chick DNA (Figure 5.11), but less dense than
the hybrid shown in Figure 5.10(b).
These results suggested that it was
possible to detect hybrid in CsZSO4 gradients by filter
hybridization but that it was not valid to assume that
the density of single-stranded ribosomal sequences
r^ras the same as that of single-stranded total chick
DNA. For example, the structural genes for 18S and
28S rRNA need not necessarily be of the same G + C
content as the spacers, and since different rnolecular
weight classes of single-stranded DNA were to be used
in the evaluation of the hybrid-density gradient pro-
cedure, it would be necessary to determine the single-
stranded density of ribosomal sequences for each size
c1ass. To achieve this, Pre-hybridization was per-
forned, the mixture centrifuged in CsrS04 gradients,
the gradients fractionated and the DNA fron the frac-
tions irnnobilized on nitrocellulose discs. The discs
ülere then split in halves, one set of half filters
being challenged with L25r 1ahe1led rRNA to detect
the hybrid and the other half being challenged with
cDNA made to rRNA to detect the single-stranded ribo-
somal sequences.
L28.
s internal single-stranded density
be accurate for a family of genes
d off the same strand of DNA. This
e the case with the ribosonal cistrons
the single-stranded density of ribo-
did not appear to vary markedly regard-
the DNA used in the pre-hybridization
kb or >/ 50 kb in length. If , however,
study were transcribed off both
the situation with the Drosophila me1-
Thi
marker can only
which are copie
would seern to b
of chick since
somal sequences
less of whether
reaction was 6
the genes under
strands, ãs is
anogaster histone genes (Lifton et ãI., L977), the
density of the sequences hybridizing to cDNA when
short DNA (slightly greater than gene length) was used
in the pre-hybridization reaction would be much lower
than that observed when DNA of repeat unit length was
used, providing al.l appropriate nRNAs l^/ere present in the
pïe-hybridization reaction. In fact, if long single-
stranded DNA were used, the hybrid peak detected by
hybridi zation using labe1led mRNAs would be expected
to be coincident with the peak detected by cDNA hybrid-
ization.
Ideally the density gradients should
L29.
contain markers for both single-stranded sequence and
100% hybrid. Thus experiments were perforrned in
which 32p labelled 100% hybrid was included in the
gradient to act as the second internal density marker.
The band.ing position of this hybrid was then deter-
rnined by Cerenkov counting of each fraction. It was
fonnd, however, that the 32p labelled cDNA, present
in the marker, bound to the nitrocellulose and rnasked
the hybrid peak being detected by hybridization withL25r labelled rRNA. varying the settings on the
scintillation counter for double 1abe11ing using L25I
and 32p did not improve the situation since there
r\rere many moïe 32p counts bound than I25I counts
hybridized. Owing to the unavailability of a y
counter, it was therefore not possible to include
both density markers in the gradients. The density
of 100% hybrid was therefore estinated in a nurnber of
independent experiments and the average of these taken
as the true densitY.
(b) Hybridi zation conditions
For the hybrid-density gradient nethod to
be of any use in deterrnining the arrangement of tan-
demly repeated genes, it was necessary that the RNA:
DNA hybridization was driven ahead of DNA reassocia-
tion during the pre-hybridization reaction and that
the DNA was not broken down during this reaction.
since the DNA being used in pre-hybridîzation was long
it was not reasonable to assume that the extent of
reassociation obtained under these conditions would
150.
be the same as that expected for molecules of length
0.5 kb at the same Cot value (0.5 kb was the length of
the DNA fragnents used in reassociation studies des-
cribed in Chapter III). Figure 5.12(a) shows that
when 20 ug each of 18S and 28S rRNA was incubated with
50 ug of single-stranded DNA of high nolecular weight
for 60 min, a considerable amount of DNA reassociation
within the ribosomal cistrons occurred. This result
conflicts with that shown in Figure 5.10 (b) where the
hybrid peak was sharp giving no evidence of reassocia-
tion within the ribosomal cistrons. The explanation
for this lies in the fact that in many of the early
pre-hybrídízation experirnents, neutralization of the
alkali denatured DNA was not performed with sufficient
care leading to precipitation of a portion of the DNA.
In these experiments vigorous agitation I^Ias used to
redissolve at least some of the DNA. It was, there-
fore, possible that less than 50 ug of DNA was used
in the experiment shown in Figure 5.10(b) and that
this DNA, at the tine of hybridization, I^Ias of some-
what smaller size than 50 kb. This alteration of
conditions may have allowed hybridization to occur
without DNA reannealing.
Figure 5.12 (b) showed that when 60 ug each
of 18S and 28S rRNA was hybridized to 50 ug of high
molecular l4leight DNA, the amount of DNA reassociation
occurring was markedly reduced. It was, therefore,
apparent that by increasing the amount of RNA in pre-
hybridization and decreasing the incubation tine, or
by decreasing the amount of DNA, alteration of the
151 .
density of the hybrid band due to DNA reassociation
hrithin the ribosomal cistrons could be avoided. The
problem of reannealing effects was expected to be
greater with DNA of length >/ 50 kb than for smaller
DNAs since the rate of reassociation is proportional
to the square root of the DNA length. Accordingly,
aLL subsequent pre-hybridi zation experiments ernployed
20 ug of a given DNA size class and 60 ug of whatever
rRNA species was neces sary .
After the initial problems with DNA
neutralization had been overcome, Figure 5.13 clearly
demonstrates that the treatments involved in the pre-
hybridization reaction induced no appreciable DNA
breakdown. Thus both the requirements of the pre-
hybri dization reaction were fulfilled.
(c) Hybrid-density studies using di fferent
single-stranded DNA size classes
For the hybrid-density method to be appli-
cable to the study of eukaryote gene arrangement, it
was necessary to deternine the relationship between
the proportion of a single-stranded DNA molecule in
hybrid form and the density of that structure relative
to single-stranded DNA and 100% hybrid. Once this
relationship is deternined for a known system, it
could be used to analyse the average gene arrangement
of any tandemly linked fanily of genes. The rela-
tionship was examined using the ribosomal cistrons of
chick as a model system. 28S rRNA was used in pre-
hybridization experiments with single-stranded DNA
r3z
size classes varying in length fron just over struc-
tural gene síze [5 kb) to repeat unit length (27 kb)
(McClenents and Skalka (1977)). In this size range
aîy given length of single-stranded DNA should have
carried" only one structural gene for 28S rRNA. By
dividing the length of the RNA by the length of the
DNA used in pre-hybridization, it was possible to
determine the average proportion of any DNA molecule
containing ribosomal sequences which should be in
hybrid form. 28S rRNA was used in this pre-hybrid-
ization step since it could be isolated in highly
purified form by phenol extraction of large ribosonal
subunits followed by rnolecular weight selection on
sucrose gradients. It was irnportant that no 18S
rRNA was present since this would have led to anornalous
hybrid densities. The proportion of the DNA nolecule
in hybrid form, âs derived by experiment, IÀlas deter-
mined by dividing the density difference between gen-
erated hybrid and single-stranded sequences by that
between 100% hybrid and single-stranded sequences.
Figure 5.14 shows the types of profiles
obtained using size classes ranging from 5.9 kb to
L4.6 kb. The sharp symmetrical peaks of cDNA hybrid-
ization, denoting the banding position of single-
stranded ribosomal sequences, suggested that DNA
reassociation within the ribosomal cistrons had not
occurred under the conditions of hybridization used.
The broad nature of the hybrid peak, however, rendered
accurate hybrid-density estinates difficult. when
r33.
the theoretical and experinental values for the per-
centage of a DNA molecule saturated with RNA were
calculated for the DNA size classes used in Figure
5.14, ãî overall trend of decreased percent saturation
with increasing DNA size was evident, but there l{Iere
few similarities in the actual estimates for aîy given
single-stranded DNA size class (Tabl.e 5.2) . It was
concluded that the relationship between these two
estimation procedures may not be a sirnple one and so
the appropriate calculations were carried out with
single-stranded DNA of many different sizes. The
theoretical percent saturation was compared with that
calculated from the experimental data for each size
class (Figure 5.15) . Fron the broad scatter of points
it was apparent that no empirical relationship could
be readily obtained from these data. This result
suggested that the basic data used for deterrnining the
experimental values for the percent saturation were not
sufficiently accurate to allow any meaningful compari-
son of these two estination procedures.
(d) E erimental shortcomi S
The hybrid-density gradient procedure
required accurate estimates for the density of 100%
hybrid, generated hybrid and single-stranded sequences.
To make accurate density estimates, the peaks needed
to be sharp and symmetrical. This was clearly not
the case for 100% hybrid (Figure 5.9(a)) and generated
hybrid (Figure 5.14) while the peaks for single-
stranded sequences seemed quite adequate. There r,tlas,
r34.
however, some variation in the single-stranded density
estimates for DNAs of different sizes (Figure 5.14),
but it was difficult to deternine whether this was
due to the different G + C contents of different
lengths of ribosomal sequence or some variation in
density estimates from gradient to gradient. This
latter type of variation could be induced by slightly
different salt concentrations (other than CsrS04) in
the different gradient buffers since the apparent
density of single-stranded DNA when loaded in T.E.
r^ras markedly different from the observed when the DNA
was loaded in pre-hybridization mix (results not shown).
To reduce the effect of such systematic errors, it
would have been best to include a 100% hybrid-density
marker in each gradient but, as described before, this
hras not possible due to the 32p 1abe1led cDNA of the
hybrid binding to nitrocellulose and rnasking the gen-
erated hybrid peak.
Broad single-stranded rnolecular weight
size classes and inaccurate size estimates could also
contribute to the observed discrepancies. The DNA
size classes, while being much better defined than
those obtained from alkaline sucrose gradient sedimen-
tation, mày sti1l have been sufficiently broad to con-
tribute to the errors observed. Such inaccurate DNA
size estimates would contribute to errors both in
experimental and theoretical estimations of the pro-
portion of the DNA nolecule in hybrid form.
Little is known about the nature of the
135.
non-transcribed spacer sequences in the chick ribo-
somal cistrons. If they l^Iere highly reiterated then
the nethods of pre-hybridization described here would
have been inadequate to drive RNA:DNA hybridization
ahead of DNA reannealing resulting in complex structures
containing hybrid, single- and double-stranded DNA.
The only evidence that DNA reassociation did not occur
comes fron the observation that the peaks of cDNA
hybridization after CsrS0O centrifugation (Figure 5.14)
'hrere sharp and symmetrical. There must, therefore,
remain some uncertainty as to whether any DNA reasso-
ciation within the ribosomal cistrons occurred during
pre-hybridization which could have contributed to the
broad hybrid peak and perhaps inaccurate hybrid-
density estimates. This particular problem, in prin-
ciple, rnight have been dispelled by carrying out pre-
hybridi zation in formarnide under conditions which per-
nitted RNA:DNA hybridízat:-on without DNA reannealing
(Casey and Davidson , I977; Vogelstein and Gillespie,
1977). Removal of formamide may have been a problen
since dialysis may have led to losses of single-
stranded DNA while ethanol precipitation may have led
to some DNA breakdown. The presence of other large
sources of error, however, indicate that this nodifi-
cation could only marginally improve the practical
estimates of the percentage of the DNA nolecule in
hybrid form.
Concludins RemarksJ
From the results presented in this chapter, it
L36.
is apparent that it was possible to isolate relatively hono-
geneous single-stranded DNA size classes by preparative
agarose ge1 electrophoresis, to detect hybrids, generated
by the pre-hybridizatîon procedure, in the density gradient
and to apparently obtain complete rRNA:DNA hybridizatio¡:
without reannealing of sequences within the ribosomal cis-
trons. However, it was not possible to ernpirically deter-
rnine the relationship between the density of the hybrid
formed, relative to those of single-stranded DNA and 100%
hybrid, and the proportion of the DNA molecule in hybrid
form. There were a nurnber of reasons for this failure,
but the main contributing factor was the inaccuracy of the
density estinates for L00% hybrid and generated hybrid.
Both of these peaks weTe broad rnaking density estimates
difficult. The 100% hybrid peak was broad because the rnole-
cules involved were small and so tended to diffuse further
around their actual banding dens ity. The breadth of the
generated hybrid peak is more difficult to explain since
the molecules involved here ltlere much longer. At least
some of the variation would be expected to derive fron the
Lact that the DNA fragrnents were generated by randon shearing
of the DNA. Thus fragments containing sequences complemen-
tary to 28S rRNA would be expected to carry a whole 28S rRNA
coding sequence or any part of it. If the RNA was not
broken down by the denaturation step prior to pre -hybridíza-
tion, those DNA fragrnents containing only a fraction of the
285 sequence would be expected to carry large portions of
single-stranded RNA rendering the structure more dense than
expected. Alternatively, if the RNA was fragrnented by
boiling, then those DNA fragments carrying a fraction of the
Is7 .
285 coding sequence r,\Iould only hybridize a fraction of the
RNA rnolecule producing a hybrid of density lower than the
average hybrid-density expected for DNA fragments of that
size. In practice, both of these effects would be expected
to be acting. This would be expected to lead to narked
over-estimates of the hybrid-density for srna11 DNA size
classes. For longer DNA nolecules, however, (about 10 kb)
the effect would be expected to be negligible since the
probability of a given DNA fragment having a complete sequence
complementary to 28S rRNA would increase with increasing
fragment size. It was, therefore, expected that the lower
rnolecular weight DNA size classes would show broader peaks
for generated hybrid than higher molecular r^reight size classes.
While a sharpening of the hybrid peak with increasing nole-
cular weight of the single-stranded DNA size class'hlas not
apparent from Figure 5.14, whenever high molecular weight
DNA was used in the pre-hybridization reaction, the hybrid
peaks obtained were quite sharp (Figures 5.10(b) and 5.L2(b).
The inabiLity to determine a relationship
between the proportion of a DNA nolecule in hybrid form and
the density of this hybrid relative to the densities of L00%
hybrid and single-stranded sequence rendered the hybrid-
density gradient procedure inadequate for determining the
average gene arrangement of a tandemly repeated gene fanily.
Although it could al1ow the determination of whether the
genes of a tandernly linked fanily are transcribed off the
same DNA strand, the large arnounts of RNA required for pre-
hybr idization and the high degree of RNA purity needed would
make the technique useful only for the study of those genes
producing a highly abundant RNA species. The currently
138.
available reconbinant DNA technology, however, permits the
same sort of inforrnation to be obtained with much more
accuracy in less time. While certain improvements could be
made to the hybrid-density gradient procedure (e.9., carry-
ing out pre-hybridizatioî under formamide conditions which
pernit RNA:DNA hybridization in the absence of DNA reanneal-
ing and the inclusion of a 100% hybrid-density marker in
each gradient, generated hybrid being detected by y counting),
it seems unlikely that the technique could yield information
about gene arrangernent of sufficient accuracy to enable
critical comparisons with gene arrangements deternined by
the recombinant DNA and Southern transfer approach.
CHAPTER VI
CONCLUDING DISCUSSION
139 .
CHAPTER VI
CONCLUDING DISCUSSION
The work described in this thesis can be divided into
two distinct sections. The significance of each part and
their linitations will be discussed separately be1ow.
Possible future research on the elucidation of the mechan-
ism of control of feather keratin genes will also be dis-
cussed.
L Structure of Feather Keratin nRNA
Fron the studies described in Chapter III and
elsewhere (Kernp, 19 75) , it is appalent that keratin rnRNA is
comprised of an heterogeneous farnily of nRNA species.
While this constitutes a very interesting system, the ina-
bility to isolate single pure nRNA species has made the
analysis of keratin nRNA structure difficult. Ultirnately,
the best way to study nRNA structure is by nucleotide
sequencing as has been done for human ß globin, rabbit ß
globin and chick ovalbumin mRNAs (Chapter I.D.1)' but with
a complex system like the avian keratin nRNAs, this can
only be achieved by cDNA cloning. Since recombinant DNA
technology has only become available to Australian bio-
chemists in the past twelve months, studies on the structure
of keratin rnRNA have had to be limited to aî analysis of
different regions of the total nRNA population using hybrid-
ization (Kernp, 1975; Kemp, Lockett and Rogers, in prepara-
tion) and DNA reassociation. Earlier studies (Kernp, 1975)
had indicated that keratin nRNA contai-ns some sequences
which are unique and others which are reiterated in the
140.
chick genome. This observation was confirmed, and in addi-
tion, by using short PolI-cDNA copied from keratin rnRNA, it
r{ras shown that at least a portion of the unique sequences
are located at the 3t end of the nRNA. By using PolI-cDNA
to prine the synthesis of longer reverse transcripts and
use of this elongated probe in reassociation kinetic analyses,
it was possible to demonstrate that unique and reiterated
sequences are covalently linked in keratin nRNA. Further-
more, density gradient analysis of reassociated duplexes
containing keratin PolI- and AMV-cDNA revealed that the 150
nucleotide sequences adjacent to the 3t poly(A) tract of
keratin nRNA, has a lower average G + C content than the
whole nRNA. From the amino acid analysis of unfractionated
feather keratin (Kenp and Rogers, 1972) ' and assuming equal
representation of each base triplet coding for a given amino
acid, it can be calculated that the keratin coding sequence
of the rnRNA should have a high G + C content (54.5%) (Lock-
ett and Kenp, 1975). The results of the density studies
on reassociated duplexes would., therefore, suggest that the
sequences adjacent to the poly(A) tract are not translated
into keratin and, therefore' constitute a 3t untranslated
sequence analogous to those found in other eukaryotic nRNAs
that have been extensively characterízed (chapter I.D.1).
Thecomplexreassociationkineticsobserved
using keratin AMV-cDNA as a probe have been explained by
postulating that genes coding for keratin (and hence the
nRNAs) contain a common sequence which is either short and
faithfully conserved, or longer and rnisnatching. This
interpretation is supported by the thermal denaturation
studies (Figure 3.8) . This is almost certainly the correct
141 .
interpretation for the naj ority of keratin genes, but the
cornplexity of the keratin nRNA population (Kenp, 1975) makes
it impossible to exclude the possibility that a subset of
the keratin genes aTe uniquely represented in the chick
genome.
Final analysis of keratin nRNA structure must
rely on sequence analysis of cloned cDNAs. Recently double-
stranded keratin cDNA has been cloned into the plasrnid
pBR322. Restriction endonuclease cleavage of genomic chick
DNA coupled with Southern transfer studies (Southern, 1975b)
using these pure probes, has ïevealed the existence of about
3 classes of nRNA. One class has been shown to reproduce
the same complex autoradiographic banding pattern as total
keratin AMV-cDNA when the filter washing procedures contain
high levels of salt. when the stringency of the washing
procedures is increased, the number of bands appearing
decreases. These results Suggest the existence of a large
fanily of cross-hybridizing genes coding for keratin. The
other classes appear to label fewer bands and the pattern
does not vary greatly with filter l^Iashing conditions
(R. B. Saint, personal communication) . These observations ,
while being in good agïeement with the results obtained by
independent nethods (chapter III), also indicate which
cloned cDNAs should be sequenced first to obtain the maxi-
mum information about keratin nRNA structure. With the
availability of cloned cDNAs, it rtlil1 also be possible to
investigate the position and nature of the reiterated
sequence in keratin nRNA. This could be achieved by caTTY-
ing out ïeassociation kinetic analysis ' or Southern transfer
studies, oû total genomic chick DNA using different restric-
r42.
tion fragments of the cDNA regions of the recombinant mole-
cules as probes.
2 Studies on Gene Organ ization
The possibility of using a density gradient
system, which exploited the density difference between RNA:
DNA hybrids and single-stranded DNA, for the study of the
arrangement of tandernly repeated genes l^ras investigated.
If the technique proved to be practical, it was to be used
to study the arrangernent of keratin'genes in chick DNA.
Since the keratin genes could only represent about 0.004%
of the chick genome, it was necessaTy to partialLy purify
these sequences to allow their easy detection in the hybrid-
density gradient procedure. The presence of contarninating
ribosomal sequences in keratin nRNA preparations also poten-
tially posed problems in the interpretation of hybrid-
density data obtained from studies involving keratin genes '
It was, therefore, desirable to obtain a quantitative
separation of DNA containing ribosomal and keratin sequences.
Initial studies were therefore caTTied out to investigate
the possibility of partialLy purifying sequences containing
keratin genes by physico-chemical rnethods '
ThernalelutionfromHAP,whilepermittingan
adequate separation of keratin and ribosornal Sequences,
resulted in substantial degradation and large losses of DNA'
Therefore, this method could not be used in the preparative
partial purification of keratin sequences (chapter IV.C.1) '
cscl gradient centrifugation using 200 ug of DNA per gradient,
allowed a quantitative separation of keratin and ribosomal
L43.
sequences presumably with minimal DNA degradation (Torniza:u¡a
and Anraku, 1965). It h-as heen shown previously that, by
fractionation of chick DNA in CsCl gradients containing
actinornycin D followed by a second CsCl step in the presence
of netropsin sulphate, an eight-fold purification of keratin
sequences was possible (Lockett and Kenp, 1975). Thus, by
using CsCl gradients, to separate ribosomal and keratin
sequences, followed by cycles of CsCl density gradient cen-
trifugation in the presence of antibiotics, a quantitative
separation of ribosonal and keratin sequences, as well as a
substantial purification of DNA containing keratin sequences 'could be achieved (Chapter IV.C.2).
The chick ribosomal cistrons were used as a
rnodel systen for the evaluation of the hybrid-density
gradient procedure as a method for the determination of gene
organization. The ïeasons for this hlere two-fold:-
(a) Ribosomal sequences constitute a suffi-
ciently large proportion of total chick
DNA to perrnit their detection without arly
prior enrichment.
(b) Pure 18S and 28S rRNA sPecies can be
easily obtained in large quantities.
Excess 28S rRNA was hybridized to different size classes of
single-stranded chick DNA under conditions which pernitted
only mininal DNA reassociation while allowing saturation of
DNA coding for 28S rRl\IA with RNA (pre-hybridization). The
resulting hybrids were centrifuged in CsTSOO gradients and
the density of hybrid deternined relative to the limiting
L44.
densities of single-stranded rihosonal sequences and 100%
hybrid. From these values, and assuming a linear relation-
ship between the proportion of the DNA in hybrid form and
the banding position of this hybrid, relative to the t\^/o
lirniting densities, an experimentally derived value for the
average percentage saturation of the DNA, containing ribo-
somal sequences, with RNA could be determined. Knowing the
single-stranded size of the DNA used in the pre-hybridizatiorL
reaction and the length of 28s rRNA, it was possible to
estirnate the average proportion of aîy given DNA rnolecule,
containing ribosomal cistrons, expected to be in hybrid
form. From this, a theoretical estimate of the percentage
of the DNA nolecule saturated with RNA could be obtained.
From the conparison of these experimentally and theoretically
derived estimates for a number of different single-stranded
DNA size classes, it was apparent, from the scatter of
points, that no meaningful relationship between these two
estimation procedures could be drawn (Figure 5.15). Thus,
the hybrid-density gradient technique appears to be insuffi-
ciently accurate for use in the analysis of the average
arrangement of tandenly repeated genes.
since the time of commencing this work, a number
of new features of gene arrangement have been revealed.
The sections below aTe designed to assess the degree of
success that an accurate hybrid-density gradient procedure
night have in detecting these arrangements.
[a) Inter en].c spacer heterog eneity
The hybrid-density gradient technique
could have , ãt best, only given information about
14s.
average gene arrangements.
size of spacers, as found in
of X. laevis and the 55
Thus, variations in the
the ribosomal cistrons
of this same toad
not have been detected.(Chapter I.E.1. (a)' (b) )
genes
could
(b) Intervening sequences ln genes
The ability to detect intervening sequences
in a tandemly linked set of genes by the hybrid-
density gradient technique, would depend on the nature
of the hybrid formed in the pre-hybridization reaction.
If each part of the structural gene sequence hybrid-
ized to a cornplete nRNA molecule, then the presence
of intervening sequences could be detected by cornparing
the densities of hybrids before and after RNAfase
treatment. Before RNAtase treatrnent, hybrids would
be expected to exhibit an anornalously high density
while this density would be rnarkedly reduced after
RNAiase treatment. If, however, one molecule of
RNA hybridized to all structural gene sequences with
out-loopings of DNA containing the intervening sequences
(as is the situation in "R-loop" mappitg), it would
not be possible to detect the intervening sequences
by the densitY Procedure.
(c) Gene rearrangements
If gene rearrangement is a prerequisite
for gene expression, the mode of detection of these
changes would be dependent on the nature of the re-
arrangements. Suppose that the inactive arrangement
of the gene farnily under study is a random distribution
L46.
around the genome while in the active arrangement,
these genes are tandernly linked. In this situation
the rearïangements could be detected by using high
molecular weight single-stranded DNA form producing
and non-producing tissues in the pre-hybridization
reaction and deternining the density of the hybrid in
CsTSOO gradients. If this mode of gene rearrangernent
occurs, hybrids forrned using DNA from non-producing
tissues would be expected to exhibit a density close
to that of single-stranded sequences while hybrids
forned using DNA fron producing tissues would show a
distinctly higher density. If, on the other hand,
fragments of the genes existed as a tandem array in
tissues where the genes t,rlere inactive, and the
remaining fragments (whether originating from another
tandem bank of fragrnents or from randorn sites in the
genome) were inserted next to the first set of frag-
ments to produce functional genes, this could be
detected by hybrid-density gradient analysis before
and after RNAtase treatment. Where gene fragments
were tandenly linked, the hybrids before RNArase
treatment would be expected to exhibit an abnornally
high density which would be observed to decrease after
RNAtase treatment. Where the whole structural gene
is present in a tandem arTay of such genes, however,
no such density difference before and after RNArase
treatment would be expected. All of these density
studies would, of course, have to be performed using
single-stranded DNA of nininum length about 10 kb
(Brown and Weber, 1968b).
r47 .
While it is clear that an accurate hybrid-
density gradient procedure rnight have been able to detect
some of the recently discovered anomalies of tandemly
repeated gene arrangements, ro details about such arrange-
ments could have been gleaned. When the arnount of DNA,
tine, effort and detail of the results obtained for the
density-gradient procedure is weighed against the same
criteria for Southern transfer analysis and recombinant DNA
techniques, it is obvious that the hybrid-density gradient
procedure, even if all üras going wel1, should not be per-
servered with.
It is apparent, therefore, that future efforts
to understand the arrangement of keratin genes in the chick
genome will rely heavily on recombinant DNA technology.
With the isolation of genomic fragments containing keratin
genes, it should be possible to definitively determine
whether the keratin genes are linked, their relationship to
each other, their polarity in the DNA and to identify, by
DNA sequencing techniques, putative elements leading to the
co-ordinate control of these genes.
Analysis of keratin gene organization, hov'rever,
can only provide linited insight into the events surrounding
the onset of terminal differentiation in the ernbryonic
feather. Ultinately it will be necessary to understand the
rnolecular events occurring in the chromatin at the time of
differentiation and the mechanism by which the dernis can
exert its control over epiderrnal differentiation.
Adans, J.M. and Cory, S. (1975). Nature 255,28-33'
Adesnik, M. and Darnell, J. E. Q972) . J ' Mol . Biol ' 9J ,
397 - 406 .
Bara1le, F.E. (I977a). Ce11 10,549-558.
Baralle, F.E. G977b). Nature 291 ,279-28I'
Baral1e, F.E. (L977c). CelI L2, 1085-1095'
Bara11e, F.E. and Brownlee, G.G. (1978). Nature 274, 84-87.
Bartels, E.A. (1945). Arch. f . Entwicklmech L42, 763-816.
Bastos, R.N. and Aviv, H. (I977). Ce11 11, 64I-650'
Beckinghan Smith, K. (1973a). Dev. Bio1. 30, 249-262'
Beckinghan srnith, K. (19i5b) . Dev. Biol . 30 , 263-278.
Beckmann, J. S . and Daniel , V. (L97 4) - J. l4o1 ' Biol ' 89 ,
35s - 362 .
Berget, S.M., Moore, C. and Sharp, P.A. (L977) ' Proc' Nat'
Acad. Sci. U.S.A. 74, 3I7I-3L75.
Berk, A.J. and sharp, P.A. (1978) . Proc. Nat. Acad. sci.
u.s.A. 75, r274-L278.
Berns, A.J.M. , Strous , G.J.A.M. and Bloemendal , H. (L972) .
Nature New Biol. 236, 7 -9.
Birnstiel, M.L., Chipchase, M. and Speirs, J' (1971) '
Progress in Nucl. Acids Res. and Mol ' Biol' 11,
351 - 389.
Birnstiel, tr4., Speirs , J., Purdom, I., Jones, K' and Leoning,
U.E. (1968) . Nature 2I9 , 454-463 '
Birnstiel, M., Telford, J., weinberg, E. and stafford, D.
(Ig74).Proc.Nat.Acad.Sci.U.S.A.7I,2900-2904.
Bonner, T.I., Brenner, D.J., Neufield, B'R' and Britten, R'J'
(1973). J. Mol. Biol. 81, 123-135'
Brack, c. and Tonegawa, s. (L977). Proc. Nat. Acad. sci.
u.s.A. 74, 5652-5656.
Bratosin, S., Horowitz, M., Laub, 0. and Aloni, Y. (1978).
Cell 13, 783-790.
Brawernann, G. (19i4) . Ann. Rev. Biochern. 42, 62I-642.
Breathnach, R" , Mandel , J. L. and Charnbon, P. (1977) .
Nature 27q, 3I4-31-9.
Britten, R.J. and Kohne, D.E. (1968). Science 161, 529-540.
Brown, D.D. and Weber, C.S. (1968a). J. Mol. Bio1. 34,
661-680.
Brown, D.D. and Weber, C. S . (1968b) . J. Mol . Biol . 3! ,
681- 697 .
Brown, D.D., Wensink, P.C. and Jordan, E. (1971). Proc.
Nat. Acad. Sci. U.S.A. 68, 3I75-31'79.
Brown, D.D., Wensink, P.C. and Jordan, E. (L972). J. Mo1 .
Biol . 63, 57 -73.
Brownlee, G.G. and Cartwright, E.M. (1976). J. Mo1. Biol-
L07 , 527 -547 .
Brownlee, G.G., Cartwright, E.M. and Brown, D.D - (L974).
J. lt4o1. Bio1. 89, 703-718-
Brownlee, G.G. Cartwright, E.I'l-, Cowan, N.J., Jarvis, J.M.
and Milstein, C. (1973). Nature New Biol. 24L,
236-240.
Brownlee, G.G. and cartwright, E.M. (L977). J. Mo1. Bio1.
LL4, 93-118.
Bukhari , A.I . , Shapiro, J.A. and Adhya, S.L. eds . (L977) '
"DNA Insertion Elenents, Plasmids and Episomes"
(New York : Cold Spring Harbour Laboratory press) '
cal1an, H.G. and Lloyd, L. (1960). Phi1. Trans. Roy, soc. B
243, L35-2r9.
Carro1l, D. and Brown, D.D. (1976a). Ce11 7, 467-475'
Carrol1, D. and Brown, D.D. (1976b). Cel1 7, 477-486'
Casey, J. and Davidson, N. (T977). Nucl . Acids Res. 4,
1539-1552.
Catteral, J.F., OtMalLey, 8.W., Robertson, M.4., Staden, R.,
Tanaka, Y. and Brownlee, G.G. (1978). Nature 271,
510- 513.
Cheng, C.C., Brownlee, G.G., Corey, N.H., Doel , M. T. ,
J. Mol. Biol . L07 ,Gi11arn, S. and Srnith, i\4. (1976) .
s27 -s45.
Chow, L.T., Gelinas, R.E., Broker, T.R.
(L977). Cell 12, 1-8.
and Roberts, R.J.
Clarkson, S . G. and Birnstiel , M. L. (19 75) . Cold Spring
Harbour Synp. Quant. Bio1. 38, 451-459.
Cohen, S. (1962). J. Biol. Chen. 237, 1555-T562.
Cohen, S. (1965) . Dev. Biol . LZ , 394-407 .
cohen, s. and carpenter, G. (1975). Proc. Nat. Acad. sci.
u.s.A. 72, T3T7-T32I.
Cohen, S. and StastnY, M. (1968).
L66, 427 -437 .
Cohn, R.H., Lowry, J.C. and Kedes,
Biochim. Biophys. Acta
9.Ce 11L.H. (1e76) .
1 47 -L6L.
Crawford, R.J. and We11s, J.R.E
3, 167-L73.
(1976) . Mo1. Bio1. RePorts
Crawford, R.J., Scott, A.C. and Wells, J.R.E. (L977) ' Eur'
J. Biochem. 72, 29L-299-
curtis, P.J. and weissrnann, c. (1976). J. Mol. Biol. 106,
1061-1075.
De Larco, J. and Guroff, G. (L973). Biochern. Biophys. Res.
Commun. 50 , 486 - 492 .
Denhardt, D.T. (1966) . Biochem. Biophys. Res. comrnun . 23,
64T-646.
Dhouailly, D. (196 7) . J . Embryol . Exptl . Morphol . !x,589-400.
Dhouailly, 0., Rogers, G.E. and Sengel, P. (1978). Dev.
Biol . 65 , 58- 68 .
Dodson, J.W. (1963) . Expt1. Ce11 Res. 3\, 233-235.
Doe1, M.T., Houghton, M., Cook, E.A. and Carey, N.H. (L977).
Nucl. Acids Res. 4, 3701-3713.
Dreyer, W.J. and Bennett, J.c. (1965). Proc. Nat. Acad. Sci.
u. s .4. 54 , 864 - 869.
Dugaiczyk, 4., Woo, S.L.C., Lai, E.C., Mace Jr., M.L.,
McReynolds, L. and O'Mal1ey, B.W. (1978). Nature
274, 328-333.
Efstratiadis , A. , Kafatos , F. C. and Maniatis , T. (I977) '
Ce11 10, 57I- 58 5.
Erikson, R. L. and Szybalski, W. (1964) . Virolo gy 2? '111 - LZ4 .
Fe11, H.B. (1957). Proc. Roy. Soc. (Biol .) 146, 242-256'
Fe11, H.B. (1962). J. Embryol. Exptl. Morphol. 10, 389-409.
Fe11, H.B. and NÍe11anby, E. (195 3) . J. Physiol . (London)
119, 470- 488.
Fitton-Jackson, F.L. and Fe11, H.B. (1963). Dev. Bio1. 7,
394-4L9.
Forget, 8.G., Marotta, C.4., Weissman, S.M., Verma, I'14',
McCaffrey, R.P. and Baltimore, D. (L974). Ann' N'Y'
Acad. Sci. 24L, 290-309.
Fraser, R.D.B., MacRae, T.P. and Rogers, G.E. (L972) '
"Keratins, Their composition, structure and Biosyn-
thesis" C.C. Thomas, Springfield, Illinois'
Furuichi, Y., l,{organ, M., }4uthukrishnan, S. and shatkin, A.J.
(1975a).Proc.Nat.Acad.Sci.U.S.A.72,362-366.
Furuichi, Y., Morgan, M., Shatkin, 4.J., Jelinek, W.,
Salditt-Georgieff, M. and Darne11, J.E. (1975b).
Proc. Nat. Acad. Sci. U.S.A. 72,1904-1908.
Garnbino, R., Kacian, D., 0rDonnell, J., Rarnirez, F. t
Marks, P.A. and Bank, A. (1974). Proc. Nat. Acad.
Sci. U.S.A. 7L, 3966-3970.
Gilbert, W. (1978). Nature 27I, 501.
Gilmore-Herbert, M. and Wal1, R. (1978). Proc. Nat. Acad.
Sci. U. S.A. 75 , 342-345.
Glover, D.M. and Hogness , D. S. (I977). Ce11 10, L67 -L76 .
Goddard, D.R. and Michaelis, L. (1954). J. Biol. Chen. 106,
605 - 614 .
Gould, M.J. and Harnlyn, P.H. (1975). FEBS Letters 50,
501- 504.
Gross, K., Probst, E. ' Schaffner, W. and Birnstiel, M.L.
(1-976a). Ce11 8' 455-469.
Gross, K. , Schaffner, W. , Telford, J. and Birnstiel, M'L'
(1976b). Ce11 8, 479-484-
Gross-Bellard, M., Oudet, P. and Chanbon, P. (T973). Eur'
J. Biochen. 36, 32-38.
Grunstein, M., Schedl, P. and Kedes, L. (1973). In "Mo1e-
cular Cytogenetics". B. Hamkalo and J. Papaconstan-
tinou eds. (New York Plenum Press) pp.1I5-T22'
Gulati, S.C., Kacian, D.L. and Spiegelrnan, S. (L974)' Proc'
Nat. Acad. Sci. U.S.A. 7T, 1035-1039.
Hanlyn, P.H., Gillhan, S., Srnith, M. and Milstein, C'
(I977). Nucl. Acdis Res. 4, \LZ3-LI34.
Harrap, B.S. and woods, E.F. (1964a) . Biochen. J. 92, 8-18.
Harrap, B. S . and Woods , E. F. (1964b) . Biochern' J ' 92 ,
19-26.
Harrap, B.S. and woods, E.F. (1967). Cornp. Biochem. Physiol .
20, 449- 460 .
Holtzer, H., Mayne, R., Weintraub, H. and Campbell, G'
(L973). In "The Biochemistry of Gene Expression in
Higher Organismsrr (Po11ak, J.K. and Lee, J.W. eds')
pp.287-304. Australia and New ZeaLand Book Co.,
Sydney.
Holtze:^, H., Sanger, J.W., Ishikawa, H. and Strahs, K'
(1972). Cold Spring Harbour Synp. Quant. Biol. 37,
549- s66.
Hoober, J.K. and cohen, s. (1967). Biochim. Biophys. Acta
138, 347-356.
Hood, L., Carnpbell, J.H. and E19in, S.C.R. (1975) ' Ann'
Rev. Genet.9,505-353.
Hood, L., Loh, 8., Hubert, J., Barstad, P., Eaton, B',
Ear\y, P., Fuhrman, J., Johnson, N., Kronenberg, M'
and schilling, J. (L976). Cold Spring Harbour synp.
Quant. Biol . 4I, 817- 836.
Hourcade, D., Dressler, D. and wolfson, J. (1973a). Proc.
Nat. Acad. Sci. U.S.A. 70, 2926-2934 '
Hourcade, D., Dressler, D. and Wolfson, J' (1975b)' Cold
Spring Harbour Synp. Quant. Bio1. 38, 537-549'
Hozumi, N. and Tonegawa, s. (1976). Proc. Nat. Acad. sci.
u.s.A. 73, 3628-3632.
Jacq, C., Mi1ler, J.R. and Brownlee, G'G' (1977) ' CeIL L2,
109-L20.
Jeffereys, A.J. and F1ave11, R.A. (L977) . Ce11 T2, 1097-1108.
Karp, R. and Hogness, D.S. (1976). Fed. Proc. 35, 1623.
Kedes, L.H. and Birnstiel, M. (1971). Nature New Bio1. 230,
165-169.
Kedes, L.H., Cohn, R., LowrY, J., Chang, A.C.Y. and Cohen,
359-370.
Kernp,
Kemp,
s.N. (197s).
D.J. (1e7s). Nature 254, 573-577.
D.J. and Rogers, G.E. (T972). Biochenistry 11,
969-975.
Kemp, D.J., Dyer, P.Y. and Rogers, G.E.
Bio1. 64, 1114-1131.
Kemp, D.J. , Partington, G.A. and Rogers,
Biochem. Biophys. Res. Commun. 60,
Kemp, D.J., Schwinghamer, M.W. and Rogers,
Mo1. Bio1. RePorts 1, 44I-446.
Kernp, D.J. , Walker, I.D. , Partington, G.A. and Rogers, G'E'
(1975). In "The Eukaryote Chronosomerr Ed.
W.J. Peacock and R.D. Brock, A.N.U. Press, Canberra'
pp.205-216.
Kersten, W., Kersten, H. and szybaLski, I4i. (1966). Biochem.
5, 236-244.
Kinniburgh,4.J., Mertz, J.E. and Ross, J. (1978). Cell 14,
681-693.
Kischer, c.w. and Furlong, N.B. (1967) . Proc. Soc. Exptl .
Bio1. Med. 124, 1188-1190.
Kitano, Y. and Kuroda, Y. (1967). Exptl. Cell Res' 48,
5s0- 560.
Kitos, P.4., Saxon, G. and Amos, H. (I972). Biochen'
Res. Commun. 47 , L426-I437 .
KlessiB, D.F. (T977). Ce11 12, 9-ZL.
Kohne, D.E. (1970). Quart. Revs. Biophys' 3, 327-375'
Lai, 8.C., Woo, S.L.C., DugaiczYk, 4., Catterall, J'F'
Biophys.
O'Mal1ey, B.W. (1978). Proc. Nat' Acad' Sci' U
75, 2205-2209.
and
S.A.
qCe 11
G
(]-974a). J. Ce11
.E. (1e74b).
1006-1014.
c . E. (I97 4c) .
Laird, C.D. (1971). Chromosoma 3J, 378-406-
Lavi, s. and Groner, Y. (L977). Proc. Nat. Acad. Sci. u.s.A.
74, 537,3--5327.
Levy, S., Childs, G. and Kedes, L. (1978). Ce11 15, ]-5l--L62.
Lifton, R.P., Goldberg, M.L., Karp, R.W. and Hogness, D'S'
(L977). Cold Spring Harbour Symp. Quant. Bio1. 42,
L047 -1052.
Lil1ie, F.R. (1965). "Li11ie's Development of the chick"
(Hanilton, H.L., ed.) pp.546-606. Ho1t, Rinehart
and Winston, New York.
Lizardi, P.M. and Brown, D.D. (1975). Ce11 4, 207-2I5'
Lockett, T.J. and Kemp, D.J. (1975) . Proc. Aust. Biochem.
Soc. 8, L07.
Mandel , J.L., Breathnach, R., Gerlinger, P., Le Meur, M',
Gannon, F. and Chambon, P. (1978). Cell 14, 64L-653'
Maniatis, T. , Kee, S. G. , Efstratiadis, A. , Kafatos , F'C'
(L976) . Ce11 8, I63-L82.
Marotta, c.4., wilson, J.T., Forget, B.G. and Weissman, s.M
(I977). J. Bio1. Chem., 252, 5040-5053.
Marushige, K., BrutIag, D. and Bonner, J. (1968). Biochem-
istry 7, 3149-3155.
Matthyssens, G. and Tonegawa, s. (1978). Nature 273, 763-
765.
Matulionis, D.H. (1970) . z. Anat. Entwickl. Gesch. I32,
L07 -L57 .
Maxam, A.M. and Gilbert, l4I. (L977). Proc. Nat. Acad. sci.
u.s.A. 74,560-564.
McClenents, I4I. and skalka, A.M. (1977). Science 196,195-
r97 .
McConauE]hY,8.L., Laird, C.D. and McCarthy, B'J' (1969)'
BiochenistrY g, 3289-3295 -
McDone1l, M.W. , Simon, M.M. and Studier, F.W. [1977) ' J '
Mo1. Biol. 119, 119-1-46.
McReynolds, L., O'MalLey, 8.W., Nisbet, 4.D., Fothergill, J.8.,
Givo1, D., Fields, S., Robertson, M. and Brownlee, G'G'
(1978). Nature 273, 723-728.
Milstein, C., Brownlee, G.G., Cartwright, 8.M., Jarvis, J'M'
and Proudfoot, N.J. (I974) - Nature 252, 354- 359 '
Modak, M.J., Marcus, S.L. and Cavalieri, L'F' (1973) '
Biochem. Biophys. Res. Commun. 55 , I-7'
Morris , c. P. and Rogers , G. E. (19 79) . Mo1 . Biol . Reports
(in press).
o'Donne11, I.J. (r973a). Aust. J. Biol. sci. 26, 401-4I3.
o'Donnell, I.J. (1975b). Aust. J. Biol. sci. 26, 415-437 -
O'Donne11, I.J. and Inglis, A.S. (L974). Aust. J. Biol' Sci'
27 , 369-382.
Partington, G.4., Kemp, D.J. and Rogers, G'E' (1973)'
Nature New Biol. 246, 33-36.
pellegrini, M., Manning, J. and Davidson, N. (L977). Cell
10, 2L3-?,24.
Ann. Rev. Biochern . 45 , 605- 629 .
Harnes , B. D. , Ke11eY, D. E. and
(19 76) . Progress in Nucl. Acids Res.
Perry,
Perry,
R. P. (1e76) .
R.P., Bard, E
Schibler, U.
and Mol. Bio1. 19, 275-29?'.
Perry, R.P., Ke11ey, D.8., Friderici, K' and Rottman, F
(1975). Ce11 4, 387-394.
Pinder, J. C. , Staynov , D,Z. and Grat zer , W' B ' G97 4) '
chenis ttY ]-3, 5373- 5378 '
Bio -
Portmann, R. , Schaffner, W. and Birnstiel, M' (1976) '
Nature 264, 3L-34.
prensky, w. (1976). Methods in cell BiologY 1é, ]-ZL-]-sz.
Proudfoot, N.J. (I976a). J. Mol. Biol . I07, 49L'SZ5'
proudfoot, N.J. (1976b) . Nuc1. Acids, Res. 3, 1811-1821.
Proudfoot, N.J. (L977). Ce11 10, 559-570'
proudfoot , N. J. and Brownlee , G. G . (L97 4) . Nature 252 '
359 - 362 .
proudfoot, N.J. and Brownlee, G.G. (1976). Nature 263,
LrL-2L4.
proudfoot, N.J. and Longley, J.I. (1976). Ce11 9, 733-746.
Proudfoot, N.J., Gillam, S., Srnith, M' and LongTey, J'I '
(L977). Ce11 11, 807-818.
Rabbitts, T.H. (1974). FEBS Letters 42, 323-326'
Rabbitts, T.H. and Forster, A. (1978). Ce11 15, 3L9-327.
Rawles, M.E. (1963). J. Ernbryol. Expt1. Morphol. 11, 765'
789.
Rawles, M.E. (1965). In "Biology of skin and Hair Growthrr.
(Lyne, A.G. and Short, B.F. eds ') pp'105-128'
Angus and Robertson, SYdneY'
RegierrJ.C.rKafatosrF.C.rKramerrK'J'rHeinrikson,R'L'and Kein, P.s. (1978a) . J. Biol . chern . 253, 1305 -1314.
Regier, J.C., Kafatos, F.C., Goodfliesh, R' and Hood, L'
(1978b).Proc.Nat.Acad.Sci.U.S.A.75,390-394.
Rochaix, J.D., Bird, A. and Bakken, A' (1974) ' J' Mol' Biol'
87, 473-487.
Romanoff, A.L. (1960). "The Avian Embryo". MacMillan'
New York.
Roop, D.R., Nordstrorn, J.L., Tsai, S'Y', Tsai, M' and
orMalLey, B.W. (1978)' Cel1 15,671-685'
Rothbêr8, s. and Eke1, T.M. cl971) . Nature 229., 34L-342.
sanger, F. and coulson, A.R. (1975). J. Mol. Biol. 94, 44L-
448 .
Schachat, F.H. and Hogness, D.S. (1973). Cold Spring Harbour
Symp. Quant. Biol. 58, 37I-381.
schaffner, I4I., Gross, K., Telford, J. and Birnstiel, M.
(1976) . Ce11 8, 47L-478-
Schaffner, W., Kunz, G., Daetwyler, H., Telford, J', Srnith,
H.O.andBirnstiel,M.L.(1978)'Ce1114,655-67L'
schultz , G. , Beato , M. and Feigelson, P . (I97 7,) . Biochen.
BiophYs. Res . Commun . 49 , 680- 689 '
Seidman , J.G., Leder, 4., Edge11, M'H', PolskL F'I', Tilgh-
man, S., Tiemeier, D.C. and L'eder, P' (1978a) ' Proc'
Nat. Acad. Sci. U.S.A. 75, 5881-5885'
Seidnan, J.G., Leder, 4., Nau, M., Norman, B' and Leder, P'
(1978b) . Science 202, LL-L7 .
Sengel, P. (1957). Experimentia 13, L77-I82'
sengel, P. (1971). Advances in Morphogenesis 9, 181-230.
Shain, W.G., Hilfer, S.R. and Fonte, V'G' (1972) ' Dev' Biol'
28, Z0z-2r8.
shine, J. and Dalgarno, L. (1975). Nature 254, 34-38.
Sinclair, J.H. and Brown, D.D. (1971) . Biochenistry 10,
276r-2769.
Snith,G.(1973).ColdSpringHarbourSyrnp.Quant.Biol.38, 507-515.
southern, E.M. (1971). Nature New Bio1. 232, 82-83.
Southern, 8.14. (1975a). J. Mol' Biol'94,51-69'
Southern, E.M. (1975b). J. Mol ' Biol' 9-8, 503-517'
Sugimoto, M. and Endo, H. (1969)' Nature 222' LZ70-I272'
Sugimoto, M., Tajirna, K., Kojina, A. and Endo, H' (I974) '
Dev. Biol . 39-, 295-307.
Sures, I., Maxarn, 4., Cohn, R.H. and Kedes, L.H' (1976)'
Ce11 9, 495-502.
Symons, R.H. (1975) . Mo1. Biol. Reports 2, 277 -285 '
szybalski, w. (1968). In "Methods in Enzymology". (Gross-
man, L. and Moldave, K., eds.) Vol. I2B pp'350-560'
Academic Press, New York.
Tartof , K.D. and Dawid, I.B. (1976). Nature 263, 27-30.
Taylor, J.M., Illmensee, R. and Summers' J. (1976)'
Biochim. Biophys. Acta 442, 324-330'
Tiemeier, D.C., Tilghman, S.M., Polsky, F.I., Seidrnan, J'G',
Leder, 4., Edge11, M.H. and Leder, P' (1978) ' Cell
L4 , 237 -245.
Tilghrnan, s.M., Tieneier, D.c., Seidman, J.G., Peterlin, 8.M.,
Sullivan, M. , MaizeL, J.V. and Leder, P' (1978a) '
Proc. Nat. Acad. Sci. U.S.A. 75,725-729'
Tilghnan, S.M., Curtis, P.J., Tiemeier, D'C', Leder, P' and
weissmann, c. (1978b). Proc. Nat. Acad. sci. u.s.A.
75,1309-1513.
Tomizawa, J. and Anraku, N. (1965). J. Mol. Biol. 11, 509-
527 .
Tonegawa, S., Hozumi, N., Matthyssens, G' and Schuller, R'
(1976).ColdSpringHarbour,Synp'Quant'Bio1'4L,
877-889.
Tonegal,\ra, S., Maxam, 4.M., Tizatd, R., Bernard, 0' and
Gilbert, W. (1978) . Proc. Nat. Acad. Sci. U.S.A. 75,
148s-1489.
vogelstein, B. and Gillespie, D. (1977). Biochen- Biophys'
Res. Commun. 75, LL27-II32 '
Vogt, V.M. (1973) . Eur. J. Biochen . 33, I9Z -200.
Voitkevich, A.A. (1966). 'tThe Feathers and Plumage of Birds"
Sidgwick and Jackson, London.
Walker, I . D. and Bridgen, J. (19 76) . Eur. J. Biochem . q,
283- 293 .
walker, I.D. and Rogers, G.E. (1976a). Eur. J. Biochern. q9,
329- 339 .
walker, I.D. and Rogers, G.E. (1976b). Eur. J. Biochem. 69,
341- 350.
14lal1ace, H. and Birnstiel, M. L. (1966) . Biochirn. Biophys .
Acta TI4, 296-3I0.
Watterson, R. L. (1942) . Physiol . ZooL. 15, 234-259 '
Wei, C.M. and Moss, B. (1975) . Proc. Nat. Acad. Sci' U'S'A'
72, 318-322.
Weinberg, E. S. , Birnstiel , M. L. , Purdom, I . F. and William-
son, R. (I972). Nature 240, 225-228.
Weinberg, E.S., Overton, G.C., Shutt, R.H. and Reeder, R'H'
(1975). Proc. Nat. Acad. Sci. U.S.A. 72,4815-4819'
Weinstock, R., Sweet, R., Weiss, M., Cedar, H. and Axel, R'
(1978). Proc. Nat. Acad. Sci. U.S.A. 75, L299-1505'
we11auer, P.K. and Dawid, I . B . (L977) . Cel1 10, I93-2I2.
We11auer, P.K., Dawid, I.8., Brown, D.D. and Reeder, R'H'
(1976a). J. Mo1. Bio1. 105, 46I-486.
We11auer, P.K., Reeder, R.M., Dawid, I.B. and Brown, D'D'
(L976b). J. Mo1. Biol. 105, 487-505'
wensink, P.c., Finnegan, 0.J., Donelson, J.E. and Hogness,
D. S . (I97 4) . Cell 3 , 315- 325 .
wesse1ls, N.K. (1961) . Expt1. Ce11 Res . 24, I3I-L42.
Wessells, N.K. (1962) . Dev. Bio1. 4, 87-L07 '
Wessells, N.K. (1965). Dev. Biol . L2, l3I-153'
white, R.L. and Hogness, D.S. [1977) . Cel1 10, 177 -L92.
Wolf, H. [1975). Analytical Biochen. 68, 505-511'
Itloodin, A.M. [1954). Biochern. J. 57,99-109'
l,tloods, E.F. (1971) . Conp. Biochen. Physiol . 59-A' 325-33L'
Wu, M. , Holrnes, D.S. , Davidson, N. r Cohn, R'H' and Kedes,
L.H. (1976) . Ce11 9' 163-169.
APPENDIX -: PUBI,ICATIONS
PAPERS PRESENTED AT MEETINGS
The organi zation of keratin genes in the chick genome. (1975)
Proc. Aust. Biochem. Soc., 8, L07 '
Unique and reiterated sequences in feather keratin nRNA'
Proc. Aust. Biochen. Soc., 11, 78'